Form N-PX Files Dataset

The Form N-PX Files Dataset contains every Form N-PX and Form N-PX/A filing submitted to SEC EDGAR from August 2003 to the present. Each record represents one complete EDGAR submission — identified by a unique accession number — and packages the filing's cover page, proxy voting record, metadata, and HTML renderings into a folder of files within a monthly ZIP container. Filers include registered management investment companies (mutual funds, ETFs, closed-end funds) and, beginning with the reporting period ending June 30, 2024, institutional investment managers subject to Section 13(f) of the Exchange Act. The dataset is the only SEC source for security-by-security, proposal-by-proposal proxy voting records at the individual fund level, covering every proxy matter on which each filer was entitled to vote during each annual reporting period (July 1 through June 30).

Update Frequency
Daily
Updated at
2026-04-29
Earliest Sample Date
2003-08-01
Total Size
10.9 GB
Total Records
127,564
Container Format
ZIP
Content Types
TXT, JSON, HTML, PDF, XML
Form Types
N-PX, N-PX/A

Dataset APIs

Programmatically retrieve the full list of dataset archive files, download URLs and dataset metadata.

Dataset Index JSON API

Download the entire dataset as a single archive file.

Download Entire Dataset:

Download a single container file (e.g. monthly archive) from the dataset.

Download Single Container:

Dataset Files

255 files · 10.9 GB
Download All
2026-04.zip140.3 KB54 records
2026-03.zip684.4 KB182 records
2026-02.zip2.3 MB126 records
2026-01.zip489.6 KB56 records
2025-12.zip4.6 MB68 records
2025-11.zip11.8 MB154 records
2025-10.zip23.3 MB157 records
2025-09.zip105.1 MB2,238 records
2025-08.zip1.4 GB26,435 records
2025-07.zip26.5 MB4,278 records
2025-06.zip62.2 KB20 records
2025-05.zip131.6 KB52 records
2025-04.zip2.0 MB62 records
2025-03.zip1.1 MB112 records
2025-02.zip462.9 KB144 records
2025-01.zip569.5 KB80 records
2024-12.zip5.0 MB100 records
2024-11.zip1.6 MB184 records
2024-10.zip6.3 MB382 records
2024-09.zip163.7 MB2,428 records
2024-08.zip1.4 GB26,809 records
2024-07.zip12.2 MB1,990 records
2024-06.zip23.1 KB3 records
2024-05.zip2.3 MB27 records
2024-04.zip75.8 KB2 records
2024-03.zip94.8 KB17 records
2024-02.zip158.9 KB1 records
2024-01.zip5.4 KB2 records
2023-12.zip1.2 MB8 records
2023-11.zip43.3 MB24 records
2023-10.zip267.3 KB5 records
2023-09.zip1.6 MB10 records
2023-08.zip486.2 MB2,530 records
2023-07.zip2.9 MB109 records
2023-05.zip8.0 KB2 records
2023-04.zip39.3 KB4 records
2023-03.zip35.1 KB4 records
2023-02.zip11.6 KB3 records
2023-01.zip12.1 KB3 records
2022-12.zip3.5 KB1 records
2022-10.zip76.2 KB6 records
2022-09.zip818.1 KB9 records
2022-08.zip454.0 MB2,518 records
2022-07.zip3.6 MB116 records
2022-06.zip206.4 KB2 records
2022-05.zip7.8 KB1 records
2022-04.zip3.1 KB1 records
2022-03.zip25.0 KB5 records
2022-02.zip2.0 KB1 records
2022-01.zip34.6 KB6 records
2021-12.zip2.1 KB1 records
2021-11.zip34.7 KB4 records
2021-10.zip51.8 KB10 records
2021-09.zip272.1 KB17 records
2021-08.zip467.5 MB2,511 records
2021-07.zip3.5 MB99 records
2021-06.zip261.0 KB6 records
2021-05.zip44.4 KB4 records
2021-04.zip381.9 KB3 records
2021-02.zip32.8 KB1 records
2021-01.zip117.9 KB3 records
2020-12.zip8.6 KB3 records
2020-11.zip3.5 MB21 records
2020-10.zip2.0 KB1 records
2020-09.zip1.9 MB22 records
2020-08.zip430.2 MB2,437 records
2020-07.zip5.6 MB126 records
2020-06.zip25.5 KB2 records
2020-05.zip572.7 KB2 records
2020-04.zip88.7 KB2 records
2020-03.zip22.6 KB3 records
2020-02.zip2.2 MB5 records
2020-01.zip170.2 KB7 records
2019-12.zip118.5 KB5 records
2019-10.zip809.3 KB5 records
2019-09.zip11.5 MB143 records
2019-08.zip399.2 MB2,417 records
2019-07.zip6.8 MB154 records
2019-06.zip7.2 KB3 records
2019-05.zip53.9 KB2 records
2019-04.zip2.7 KB1 records
2019-03.zip18.2 KB6 records
2019-02.zip6.3 KB1 records
2019-01.zip691.6 KB4 records
2018-11.zip11.6 KB3 records
2018-10.zip13.1 KB4 records
2018-09.zip3.2 MB12 records
2018-08.zip456.6 MB2,648 records
2018-07.zip7.0 MB144 records
2018-06.zip394.2 KB2 records
2018-05.zip32.1 KB6 records
2018-04.zip2.2 KB1 records
2018-03.zip16.7 KB4 records
2018-02.zip10.6 KB3 records
2018-01.zip12.7 KB4 records
2017-12.zip388.9 KB3 records
2017-11.zip48.2 KB7 records
2017-10.zip863.2 KB3 records
2017-09.zip1.5 MB18 records
2017-08.zip446.0 MB2,685 records
2017-07.zip3.5 MB135 records
2017-06.zip28.0 KB2 records
2017-05.zip35.4 KB3 records
2017-04.zip6.6 KB1 records
2017-03.zip1.9 KB1 records
2017-02.zip145.5 KB7 records
2017-01.zip17.1 KB5 records
2016-12.zip2.2 MB15 records
2016-11.zip1.8 MB13 records
2016-10.zip193.5 KB4 records
2016-09.zip676.8 KB11 records
2016-08.zip421.0 MB2,690 records
2016-07.zip4.1 MB133 records
2016-06.zip773.8 KB4 records
2016-05.zip11.7 KB2 records
2016-04.zip642.6 KB8 records
2016-03.zip2.2 MB2 records
2016-02.zip32.7 KB5 records
2016-01.zip1.2 MB3 records
2015-12.zip88.5 KB5 records
2015-11.zip685.0 KB2 records
2015-10.zip8.5 MB9 records
2015-09.zip6.5 MB21 records
2015-08.zip403.2 MB2,731 records
2015-07.zip6.4 MB149 records
2015-06.zip3.0 KB1 records
2015-05.zip88.5 KB8 records
2015-04.zip21.2 KB1 records
2015-03.zip51.9 KB5 records
2015-02.zip29.6 KB2 records
2015-01.zip17.7 KB4 records
2014-12.zip9.7 KB4 records
2014-11.zip600.7 KB5 records
2014-10.zip2.2 MB4 records
2014-09.zip961.5 KB11 records
2014-08.zip388.2 MB2,650 records
2014-07.zip7.1 MB176 records
2014-06.zip120.5 KB15 records
2014-05.zip52.2 KB2 records
2014-04.zip18.1 KB3 records
2014-03.zip232.5 KB6 records
2014-02.zip10.5 KB1 records
2014-01.zip4.6 MB3 records
2013-12.zip826.0 KB46 records
2013-11.zip54.3 KB10 records
2013-10.zip359.1 KB5 records
2013-09.zip3.0 MB32 records
2013-08.zip363.4 MB2,640 records
2013-07.zip6.1 MB181 records
2013-06.zip199.2 KB6 records
2013-05.zip8.1 MB17 records
2013-04.zip3.8 KB1 records
2013-03.zip2.7 KB1 records
2013-02.zip9.4 KB3 records
2013-01.zip1.8 MB7 records
2012-12.zip92.0 KB4 records
2012-10.zip345.5 KB8 records
2012-09.zip2.5 MB38 records
2012-08.zip425.8 MB2,702 records
2012-07.zip2.4 MB147 records
2012-06.zip131.4 KB2 records
2012-05.zip5.6 KB1 records
2012-04.zip180.6 KB5 records
2012-03.zip59.3 KB4 records
2012-01.zip2.5 KB1 records
2011-12.zip20.1 KB6 records
2011-11.zip524.1 KB2 records
2011-10.zip12.8 MB27 records
2011-09.zip1.2 MB31 records
2011-08.zip415.3 MB2,739 records
2011-07.zip2.6 MB131 records
2011-06.zip1.8 KB1 records
2011-05.zip85.7 KB2 records
2011-04.zip367.1 KB19 records
2011-03.zip6.0 KB1 records
2011-02.zip23.6 KB5 records
2011-01.zip24.4 KB1 records
2010-12.zip6.5 KB1 records
2010-11.zip5.5 KB2 records
2010-10.zip479.2 KB7 records
2010-09.zip7.2 MB27 records
2010-08.zip361.5 MB2,950 records
2010-07.zip2.4 MB133 records
2010-06.zip39.6 KB4 records
2010-05.zip10.8 KB2 records
2010-04.zip173.1 KB16 records
2010-03.zip1.6 MB7 records
2010-02.zip37.8 KB2 records
2010-01.zip3.3 KB1 records
2009-12.zip295.2 KB4 records
2009-11.zip115.9 KB5 records
2009-10.zip3.4 MB8 records
2009-09.zip5.7 MB22 records
2009-08.zip305.2 MB2,871 records
2009-07.zip3.2 MB176 records
2009-06.zip59.8 KB2 records
2009-05.zip930.2 KB4 records
2009-04.zip202.5 KB3 records
2009-03.zip185.2 KB1 records
2009-02.zip59.2 KB4 records
2009-01.zip28.5 KB2 records
2008-12.zip1.9 MB7 records
2008-11.zip65.1 KB5 records
2008-10.zip2.5 MB36 records
2008-09.zip16.8 MB91 records
2008-08.zip282.3 MB3,301 records
2008-07.zip1.2 MB123 records
2008-05.zip138.9 KB3 records
2008-04.zip36.3 KB7 records
2008-03.zip97.0 KB1 records
2008-02.zip74.6 KB1 records
2008-01.zip1.0 MB8 records
2007-12.zip217.1 KB3 records
2007-11.zip983.0 KB4 records
2007-10.zip2.9 MB6 records
2007-09.zip4.1 MB32 records
2007-08.zip257.4 MB3,331 records
2007-07.zip1.0 MB122 records
2007-06.zip2.1 MB11 records
2007-05.zip2.6 MB7 records
2007-04.zip10.9 KB3 records
2007-03.zip295.9 KB7 records
2007-02.zip630.0 KB10 records
2007-01.zip3.0 KB1 records
2006-12.zip22.5 KB6 records
2006-11.zip6.9 MB18 records
2006-10.zip3.0 MB8 records
2006-09.zip7.4 MB44 records
2006-08.zip212.5 MB3,203 records
2006-07.zip2.2 MB221 records
2006-06.zip184.1 KB10 records
2006-05.zip380.9 KB5 records
2006-04.zip933.7 KB3 records
2006-03.zip499.0 KB5 records
2006-02.zip106.9 KB4 records
2006-01.zip1.7 MB14 records
2005-12.zip6.4 MB51 records
2005-11.zip4.6 MB48 records
2005-10.zip176.3 KB11 records
2005-09.zip2.6 MB34 records
2005-08.zip200.7 MB3,365 records
2005-07.zip1.1 MB288 records
2005-06.zip430.9 KB14 records
2005-05.zip6.5 KB3 records
2005-04.zip769.9 KB10 records
2005-03.zip368.9 KB14 records
2005-02.zip150.2 KB60 records
2005-01.zip3.0 MB85 records
2004-12.zip1.6 MB17 records
2004-11.zip7.1 MB87 records
2004-10.zip3.5 MB74 records
2004-09.zip15.4 MB119 records
2004-08.zip195.7 MB3,552 records
2004-07.zip208.0 KB19 records
2003-08.zip22.8 KB1 records

What This Dataset Contains

Form N-PX is an annual report of proxy voting records introduced in 2003 under Section 30 of the Investment Company Act of 1940. The dataset spans all N-PX and N-PX/A submissions from August 2003 to the present. Each record is one complete EDGAR submission for a Form N-PX or Form N-PX/A filing, stored as a folder of files within a monthly ZIP container. The unit of observation is the filing, not an individual proxy vote — a single record may contain thousands of vote-level entries embedded in its proxy voting record document, or it may contain no vote data at all if the filer had no votes to report.

The dataset contains two fundamentally different record shapes reflecting a structural discontinuity. Filings from 2003 through roughly 2023 were submitted in heterogeneous formats (plain text, HTML, PDF), often combining cover page and voting records into a single monolithic document. Beginning with the reporting period ending June 30, 2024, the SEC mandated structured XML using SEC-defined schemas, producing a uniform two-document XML structure with consistently named and typed fields. Users working across the full historical span must account for this format transition.

Content Structure of a Single Record

Folder Structure and File Inventory

Each record is a folder named with an 18-digit zero-padded accession number (dashes removed). For example, folder 000180755925000005 corresponds to accession number 0001807559-25-000005. A record in the structured XML era contains up to five files:

  1. metadata.json — always present; dataset-provider-generated filing metadata (not part of the original EDGAR submission).
  2. primary_doc.xml — the main N-PX submission document: cover page, filer identification, and signature.
  3. xslN-PX_X01/primary_doc.xml — an XSL-transformed XHTML rendering of the primary document.
  4. A proxy voting record XML file (variable filename, e.g., VoteTable.xml, N-PX2024.xml, SE-CN-PX-Sample2.xml) — the tabular proxy voting data. Present only when the filer actually cast votes.
  5. xslNPX-INFO-TABLE_X01/<filename>.xml — an XSL-transformed XHTML rendering of the proxy voting record. Present only when the voting record file exists.

Notice-only filings — where the filer reports no votes — contain only the primary document, its XHTML rendering, and the metadata file; no voting record file or its rendering is present.

For pre-2024 filings, the substantive documents may be HTML, plain text, or PDF rather than XML, and the cover page and voting records may be combined into a single monolithic document rather than separated into distinct files. The metadata.json and folder structure remain consistent across all eras.

Component-by-Component Breakdown

metadata.json

A JSON object consolidating EDGAR index information and direct URLs. Key fields:

  • formType: "N-PX" or "N-PX/A".
  • accessionNo: Dash-separated accession number (e.g., "0001807559-25-000005").
  • filedAt: Filing timestamp with timezone offset.
  • periodOfReport: End date of the reporting period, typically YYYY-06-30.
  • effectivenessDate: Date the filing became effective on EDGAR.
  • description: Human-readable form description.
  • id: Hexadecimal record identifier assigned by the dataset provider.
  • entities: Array of filer entity objects. Each contains the company name (with a role annotation such as "Filer"), CIK, filing type, the act under which the filing is made ("40" for the Investment Company Act, "34" for the Exchange Act), SEC file number, IRS number, state of incorporation code, fiscal year end (MMDD format), and film number. A filing may have multiple entity entries when it is made on behalf of multiple registrants or involves both an investment company and its manager.
  • documentFormatFiles: Array describing every document in the submission. Each entry specifies a sequence number, document type (e.g., "N-PX" or "PROXY VOTING RECORD"), file size in bytes, and a direct URL to the document on SEC.gov. XSL-rendered versions appear as separate entries with blank size values; the underlying XML source files carry the actual byte sizes.
  • dataFiles: Array of supplemental data files; typically empty for N-PX filings.
  • seriesAndClassesContractsInformation: Array for series and class identifiers; typically empty.
  • linkToFilingDetails, linkToTxt, linkToHtml: URLs to the EDGAR filing detail page, complete submission text file, and index page.
  • linkToXbrl: Typically empty; N-PX uses SEC-specific XML schemas, not XBRL taxonomies.

primary_doc.xml (Cover Page and Signature)

A structured XML document in the http://www.sec.gov/edgar/npx namespace (with the http://www.sec.gov/edgar/common namespace for address elements). It contains two principal blocks:

headerData encodes submission-level metadata:

  • submissionType: "N-PX" or "N-PX/A".
  • filerInfo: registrant type (IC for investment company, IM for institutional manager), CIK, CCC (masked), live-or-test flag, investment company type (often empty for managers), and the period of report in MM/DD/YYYY format.

formData contains three sub-elements:

  • coverPage: The registrant's legal name, phone number, mailing address, report type designation, calendar year, year-or-quarter indicator, file number, and a confidential treatment election flag. Optional elements include agentForService (name and address of the agent for service of process), reportingCrdNumber (CRD number for investment advisers), and reportingSecFileNumber (SEC file number distinct from the EDGAR file number). The reportInfo block is critical: its reportType field distinguishes filings with actual voting data (e.g., INSTITUTIONAL MANAGER VOTING REPORT, REGISTRANT VOTING REPORT) from notice-only filings (INSTITUTIONAL MANAGER NOTICE REPORT, REGISTRANT NOTICE REPORT). Notice-only filings include a noticeExplanation sub-element with a reason such as "REPORTING PERSON HAS POLICY TO NOT VOTE". The explanatoryInformation block indicates whether the filer has appended explanatory notes.
  • summaryPage (optional): Contains otherIncludedManagersCount, indicating how many additional managers' voting records are aggregated into the filing.
  • signaturePage: The reporting person's name, the electronic signature (txSignature), the printed name (txPrintedSignature), the signer's title, and the signature date.

The primary document does not contain proxy voting records. Those reside in a separate file.

Proxy Voting Record XML (e.g., VoteTable.xml)

The substantive core of any filing that reports actual votes. This is a structured XML document in the http://www.sec.gov/edgar/document/npxproxy/informationtable namespace with a proxyVoteTable root element containing a flat sequence of proxyTable entries. Each proxyTable entry represents one matter voted upon at one shareholder meeting for one issuer. Fields per entry:

  • issuerName: Full legal name of the company whose shares were voted.
  • cusip: Nine-character CUSIP identifier.
  • isin: Twelve-character ISIN, providing an international complement to CUSIP.
  • meetingDate: Shareholder meeting date in MM/DD/YYYY format.
  • voteDescription: Free-text description of the matter voted upon, ranging from short labels ("Elect Director John Smith") to full paragraph-length shareholder proposal summaries.
  • voteCategories: Nested element containing one or more voteCategory sub-elements, each with a categoryType string. Common values include DIRECTOR ELECTIONS, SECTION 14A SAY-ON-PAY VOTES, RATIFICATION OF AUDITOR, SHAREHOLDER PROPOSAL, and other governance classifications.
  • voteSource: ISSUER for management proposals, SHAREHOLDER for shareholder-initiated proposals.
  • sharesVoted: Total shares voted on this matter. A value of 0 means the filer held the security but did not cast a vote.
  • sharesOnLoan: Shares on loan at the record date and unavailable for voting.
  • vote (conditional): Present only when sharesVoted is greater than zero. Contains one or more voteRecord sub-elements, each specifying:
    • howVoted: The vote direction — FOR, AGAINST, ABSTAIN, WITHHOLD, 1 YEAR, 2 YEARS, 3 YEARS, or other values.
    • sharesVoted: Number of shares voted in this direction.
    • managementRecommendation: What management recommended (e.g., FOR, AGAINST).

Multiple voteRecord elements within a single proxyTable entry indicate the filer split its shares across different vote directions on the same matter. A single filing can contain anywhere from a few entries to tens of thousands, depending on the breadth of the filer's portfolio.

XSL-Transformed HTML Renderings

The subdirectories xslN-PX_X01/ and xslNPX-INFO-TABLE_X01/ contain XHTML renderings of the primary document and the proxy voting record, respectively. These are visual representations of the same structured XML data, formatted with SEC-standard styling (HTML tables, form-like layouts, radio-button indicators for vote choices). They replicate the browser-rendered view from the EDGAR filing viewer and contain no additional data beyond the source XML. They exist as convenience renderings.

Included and Excluded Content

Each dataset record includes the metadata.json file and all documents from the original EDGAR submission except image files. For structured-era filings, this means the primary document XML, the voting record XML (if applicable), and the XHTML renderings. For older filings, this means whatever text, HTML, or PDF documents composed the original submission.

Image files (logos, graphical elements) present in the original EDGAR submission are excluded. XBRL taxonomy files are not applicable since N-PX uses SEC-specific XML schemas. The complete submission text file referenced in linkToTxt is an EDGAR-generated concatenation of all documents and is not separately stored in the dataset folder — it is accessible via URL only.

Structural Evolution Over Time

2003 through roughly 2023 (unstructured era): Filers submitted proxy voting records in free-form text, HTML tables, or PDF. The cover page and voting records were often combined into a single document. There was no standardized field naming, column ordering, or delimiter convention. CUSIP and other security identifiers were inconsistently reported; many filings used only issuer names or ticker symbols. Machine parsing of these historical records requires filing-specific or heuristic-based extraction logic.

2024-present (structured XML era): The SEC mandated XML schemas impose uniform structure. Fields are consistently named, typed, and ordered. The filing is separated into a primary document and a distinct voting record file. CUSIP and ISIN are mandatory. New fields include sharesOnLoan, voteCategories, and quantitative per-direction share counts. XSL-transformed XHTML renderings are generated automatically by EDGAR and included alongside the source XML.

Interpretation and Extraction Notes

Notice-only filings: A record with reportType set to a notice designation (e.g., INSTITUTIONAL MANAGER NOTICE REPORT) contains no proxy voting data. The noticeExplanation element in the cover page states the reason, such as a policy of not voting proxies or holding no voteable securities during the period. These records are materially smaller.

Vote element absence versus abstention: A proxyTable entry with sharesVoted of zero lacks the vote sub-element entirely. This represents a held position where no vote was cast and is distinct from ABSTAIN, which is an affirmative voting action recorded in a voteRecord.

File naming variability: The proxy voting record XML file has no fixed filename. Filer-chosen names vary (e.g., VoteTable.xml, N-PX2024.xml, SE-CN-PX-Sample2.xml). To locate it programmatically, filter the documentFormatFiles array in metadata.json for entries with type "PROXY VOTING RECORD".

Document sequence and XSL pairs: In the documentFormatFiles array, the XSL-rendered version and the source XML for the same document share a sequence number. The rendered version has a blank size field and a URL path containing the XSL subdirectory; the source XML has an actual byte size.

Multiple entities: The entities array may contain multiple entries when a filing covers multiple registrants or involves both an investment company and its manager. The act field ("40" or "34") indicates the statutory basis for the filing obligation.

Schema version comment: The proxy voting record XML may include an XML comment indicating the schema version (e.g., <!--VoteTableSchemaVersion:X0300-->), which can be useful for tracking format changes across filing cycles.

Historical format heterogeneity: For pre-2024 filings, document files may be HTML, plain text, or PDF rather than XML. Some early filings contain a single monolithic document combining cover page, voting records, and signature. The metadata.json and folder structure remain consistent, but substantive content files require format-specific parsing strategies.

Who Files or Publishes This Dataset, and When

Who files

The primary filer population is registered management investment companies under the Investment Company Act of 1940. This includes:

  • Open-end management investment companies (mutual funds), including money market funds that hold voting securities.
  • Exchange-traded funds (ETFs) organized as management investment companies (not unit investment trusts).
  • Closed-end management investment companies, including interval funds and tender offer funds.

A single filing typically covers an entire fund complex. The EDGAR registrant is the legal trust or corporation that sponsors the fund family, and the filing reports proxy votes for all series (individual fund portfolios) under that registrant.

Expanded filer population (reporting periods ending June 30, 2024, onward). The SEC's 2022 amendments to Form N-PX added two new filer classes:

  • Institutional investment managers that file Form 13F must now report their say-on-pay votes (executive compensation advisory votes under Exchange Act Section 14A) on Form N-PX, under new Rule 14Ad-1. These filings appear in the dataset beginning August 2024.
  • Business development companies (BDCs), which elect regulation under parts of the 1940 Act but were previously exempt from N-PX, are now also required to file.

Who does not file

  • Unit investment trusts (UITs) — not management investment companies, so exempt from N-PX.
  • Foreign private issuers and operating companies — file proxy materials on DEF 14A and related forms, not N-PX.
  • Unregistered pooled vehicles (private funds, hedge funds) — outside the 1940 Act registration regime.

Reporting trigger and schedule

Form N-PX is a periodic, calendar-driven filing, not event-driven.

ElementDetail
Reporting periodJuly 1 through June 30 each year
Filing deadlineAugust 31 each year
TriggerAnnual obligation; no variation by filer size, accelerated status, or fiscal year
CoverageEvery proxy matter on which the filer was entitled to vote or voted during the period

Every registrant that held any voting securities during the reporting period must file. A fund that held no voting securities may file with a minimal or empty record.

Amendments (N-PX/A)

A filer may submit Form N-PX/A at any time to correct or amend a prior filing. There is no separate amendment deadline. Amendments fully supersede the original for the same reporting period. The dataset includes both originals and amendments as separate records, each with its own accession number.

Earliest records

The SEC adopted Form N-PX in January 2003 under Rule 30b1-4 of the 1940 Act. The first reporting period covered votes cast during the twelve months ending June 30, 2003, with filings due by August 31, 2003. The earliest records in the dataset date from August 2003. Form N-PX was created for electronic EDGAR filing; there is no pre-EDGAR paper filing history.

Key distinctions

DimensionDetail
Governing rulesRule 30b1-4 (1940 Act) for funds; Rule 14Ad-1 (Exchange Act) for 13F managers
Amendment formN-PX/A, filed as needed
Nearby forms, different regimeDEF 14A (operating-company proxy statements); Form 13F (institutional holdings, separate from voting); N-CSR (fund shareholder reports)

How This Dataset Differs From Similar Datasets or Filings

Form N-PX is the only SEC filing dedicated to recording how registered investment companies voted their proxied shares. The most relevant neighboring datasets fall into three groups: other fund-level periodic reports under the Investment Company Act, the proxy materials that generate the votes N-PX records, and issuer-side vote result disclosures.

Form N-PORT — Monthly/Quarterly Portfolio Holdings N-PORT is the closest sibling. Both are filed by registered management investment companies and produce structured, security-level data. The split is positional versus behavioral: N-PORT reports what a fund holds and the financial characteristics of those positions; N-PX reports how a fund voted on ballot items at portfolio companies. N-PORT is filed monthly (public quarterly); N-PX is filed once annually by August 31, covering the twelve-month period through June 30. The two can be joined on fund and security to connect holdings with voting decisions, but neither substitutes for the other.

DEF 14A / PRE 14A — Proxy Statements Proxy statements are upstream of N-PX. A DEF 14A filed by an operating company describes the matters to be voted on at a shareholder meeting — director elections, compensation plans, shareholder proposals. N-PX then records how each fund voted on those matters. The filer populations do not overlap: DEF 14A is filed by the company soliciting votes; N-PX is filed by the fund casting votes. Proxy statements contain full proposal text, board recommendations, and supporting arguments; N-PX contains only a brief matter description and the fund's vote. Linking the two requires matching on issuer, meeting date, and proposal.

Form 8-K Item 5.07 — Voting Results After a shareholder meeting, operating companies disclose aggregate vote tallies under 8-K Item 5.07. This covers the same underlying events as N-PX but from opposite sides: 8-K Item 5.07 gives the total count from the issuer's perspective; N-PX gives each fund's individual vote from the investor's perspective. Aggregate outcome analysis requires 8-K; attribution of votes to specific institutional holders requires N-PX.

Form N-CSR / N-CSRS — Certified Shareholder Reports N-CSR (annual) and N-CSRS (semi-annual) are the principal narrative periodic reports for registered investment companies, analogous to 10-K/10-Q for operating companies. They contain financial statements, management discussion, and portfolio schedules. The overlap with N-PX is limited to sharing the same filer population. N-CSR covers financial performance and portfolio composition; N-PX covers proxy voting behavior exclusively. There is no content overlap.

Form 13F — Quarterly Institutional Holdings 13F is filed by institutional investment managers with at least $100 million in Section 13(f) securities. It partially overlaps with N-PX in filer population — mutual funds file both, but hedge funds file 13F without filing N-PX. The legal regimes differ: 13F arises under the Exchange Act; N-PX under the Investment Company Act. Like N-PORT, 13F reports holdings positions, not voting decisions, so it complements rather than duplicates N-PX.

Form N-PX/A — Amendments N-PX/A filings are amendments that correct or supplement a previously filed N-PX. They follow the same structure and content requirements. When an N-PX/A exists for a given fund-year, it generally supersedes the original. Both N-PX and N-PX/A are included in this dataset.

What Makes This Dataset Distinct

No other SEC filing captures security-by-security, proposal-by-proposal proxy voting records at the individual fund level. Holdings datasets (N-PORT, 13F) show what funds own but not how they vote. Proxy statements (DEF 14A) show what is being voted on but not how any fund voted. Voting results (8-K Item 5.07) show aggregate outcomes but not individual institutional votes. N-PX is the sole source for reconstructing the governance decisions of specific registered investment companies across their full range of portfolio holdings.

Who Uses This Dataset

Form N-PX filings record how mutual funds and ETFs voted on every proxy ballot item during the prior year, at the level of individual issuer, meeting, and proposal. This ballot-level granularity serves distinct professional workflows.

Stewardship and governance analysts

Governance teams at asset managers and asset owners benchmark proxy voting across fund families and over time. They compare votes on identical ballot items — say-on-pay, board elections, shareholder resolutions on climate or executive compensation — against stated stewardship principles. Key fields: matter description, management recommendation, and recorded vote. Output includes peer-comparison dashboards, engagement priority lists, and voting-guideline revisions.

ESG and responsible-investment researchers

Researchers at ESG rating firms, responsible-investment consultancies, and pension advisory teams score fund managers on support for environmental, social, and governance proposals. They filter matter descriptions for climate disclosure, diversity reporting, and lobbying transparency, then aggregate vote outcomes by fund complex. The resulting scorecards help institutional allocators verify that a manager's voting record matches its sustainability commitments.

Proxy advisory and vote-analytics teams

Proxy advisory firms and independent vote-analytics providers ingest N-PX filings at scale to build historical voting databases. They parse each ballot item's description, the fund's vote, and the management recommendation to compute pass rates, dissent levels, and swing-vote dynamics. These analytics power client voting-decision platforms and annual voting-trend reports.

Activist monitoring and corporate defense advisors

Corporate advisory practices at investment banks and law firms track N-PX data to predict which fund complexes will support or oppose dissident slates, governance reforms, and merger proposals. They focus on contested director elections, proxy fights, and M&A votes. This intelligence informs board-level engagement strategy and settlement negotiations with activist investors.

Securities and fund-disclosure counsel

Securities attorneys review N-PX filings for compliance with SEC reporting requirements — checking completeness of matter descriptions, accuracy of vote classifications, and consistency between original and amended (N-PX/A) filings. They also benchmark disclosure practices across peer registrants to support compliance opinions and responses to SEC comment letters.

Fund compliance teams

Compliance officers at fund companies cross-reference N-PX vote records against internal order logs and proxy-execution confirmations to verify that recorded votes match firm voting policies and client instructions. Amended filings (N-PX/A) flag prior reporting errors requiring root-cause analysis and are a key audit artifact.

Quantitative and event-driven researchers

Systematic strategy teams link ballot-level votes to security identifiers (CUSIP) and meeting dates to test whether institutional voting patterns predict stock returns, corporate actions, or governance changes. The vote outcome, security identifier, and meeting date are the critical fields. These signals feed factor models and event-study pipelines.

Financial data engineers and NLP teams

Data engineering teams build pipelines to parse, normalize, and structure N-PX filings across heterogeneous formats — XML, HTML, TXT, JSON, and PDF — spanning two decades of EDGAR submissions. They extract structured fields (issuer, CUSIP, matter description, vote cast, management recommendation) into a consistent schema. NLP teams use matter-description text to train ballot-item classifiers (compensation, environmental, board composition) and to build retrieval systems that answer questions about fund voting history.

Data journalists and governance advocacy organizations

Journalists at financial news outlets and analysts at governance-focused advocacy organizations aggregate N-PX votes by fund complex across high-profile categories — climate resolutions, CEO pay, political-spending disclosure — and publish annual scorecards ranking the largest fund families by support levels. The matter description and vote-cast fields drive this work, and the resulting reports shape public discourse on institutional investor accountability.

Specific Use Cases

Benchmarking fund-family voting alignment on ESG shareholder proposals

Filter the proxy voting record by voteCategories values such as SHAREHOLDER PROPOSAL and voteSource of SHAREHOLDER, then isolate matter descriptions referencing climate disclosure, political spending, or diversity reporting. Aggregate each fund complex's support rate across these categories and compare against stated stewardship commitments. The output is a scorecard ranking fund families by ESG proposal support, used by pension consultants and allocators to verify manager alignment.

Detecting dissent on say-on-pay votes across large fund complexes

Extract all ballot entries where voteCategories includes SECTION 14A SAY-ON-PAY VOTES and compare howVoted against managementRecommendation. Identify funds that voted against management on compensation matters and track dissent rates over multiple filing years. Corporate advisory teams use these patterns to anticipate opposition before annual meetings and to prioritize board-level engagement with specific fund families.

Building a historical proxy voting database across format eras

Ingest the full archive from 2003 to the present, handling the structural discontinuity between pre-2024 free-form filings (HTML, text, PDF) and post-2024 schema-validated XML. Use the documentFormatFiles array in metadata.json to locate proxy voting record documents by type, then apply format-specific parsers for older filings and XML extraction for structured ones. The resulting normalized database — keyed on fund CIK, CUSIP, meeting date, and matter description — powers vote-analytics platforms and historical trend reports.

Linking ballot-level votes to stock returns for event-study research

Join proxy voting records to equity price data using the CUSIP field and meeting date. Isolate contested director elections or anti-management votes where large fund complexes broke from the management recommendation, then measure abnormal returns around the meeting date. Quantitative researchers use these signals to test whether institutional voting patterns carry predictive information for governance-driven price moves.

Monitoring amended filings to flag reporting corrections

Track N-PX/A submissions by filtering metadata.json for formType of N-PX/A. Compare the amended voting record against the original filing for the same fund and reporting period to identify changed vote entries — altered vote directions, corrected share counts, or added ballot items. Fund compliance teams use these comparisons for root-cause analysis, and securities counsel use amendment frequency as a peer benchmark for disclosure quality.

Mapping institutional voting coalitions on contested proxy fights

For a specific issuer and meeting date, collect all N-PX ballot entries across every filing fund. Group funds by vote direction on contested matters — dissident director slates, merger proposals, governance reforms — and identify which fund complexes voted together. Activist advisors and corporate defense teams use these coalition maps to model likely vote outcomes and inform settlement or engagement strategy ahead of future contests.

Dataset Access

Dataset Index JSON API: https://api.sec-api.io/datasets/form-npx-files.json

This endpoint returns metadata about the Form N-PX Files Dataset, including the dataset name, description, last updated timestamp, earliest sample date, total records, total size, covered form types, container format, and file types. It also returns the download URL for the entire dataset and a list of all individual container files with per-container metadata such as size, record count, last updated timestamp, and download URL. No API key is required to access this endpoint.

Use this endpoint to monitor which containers have been updated in the most recent refresh run, so you can selectively download only the containers that changed on a given day.

Example
1 {
2 "datasetId": "1f13365b-9ae0-68f2-9aa3-786115639911",
3 "datasetDownloadUrl": "https://api.sec-api.io/datasets/form-npx-files.zip",
4 "name": "Form N-PX Files Dataset",
5 "updatedAt": "2026-04-17T02:59:06.305Z",
6 "earliestSampleDate": "2003-08-01",
7 "totalRecords": 127530,
8 "totalSize": 10946619715,
9 "formTypes": ["N-PX", "N-PX/A"],
10 "containerFormat": "ZIP",
11 "fileTypes": ["TXT", "JSON", "HTML", "PDF", "XML"],
12 "containers": [
13 {
14 "downloadUrl": "https://api.sec-api.io/datasets/form-npx-files/2026/2026-03.zip",
15 "key": "2026/2026-03.zip",
16 "size": 13818783,
17 "records": 154,
18 "updatedAt": "2026-03-21T02:51:19.000Z"
19 }
20 ]
21 }

Download Entire Dataset: https://api.sec-api.io/datasets/form-npx-files.zip?token=YOUR_API_KEY

Downloads the full Form N-PX Files Dataset as a single ZIP archive containing all containers. This endpoint requires an API key passed via the token query parameter.

Download Single Container: https://api.sec-api.io/datasets/form-npx-files/2026/2026-03.zip?token=YOUR_API_KEY

Downloads an individual monthly container file instead of the full dataset. Replace the year and month path segments with the desired period. This endpoint requires an API key passed via the token query parameter.

Frequently Asked Questions

What form does this dataset cover?

The Form N-PX Files Dataset covers Form N-PX and its amendment variant Form N-PX/A. Form N-PX is the annual report of proxy voting records filed with the SEC under Section 30 of the Investment Company Act of 1940.

What does one record in this dataset represent?

Each record represents one complete EDGAR submission for a Form N-PX or N-PX/A filing, identified by a unique accession number. The unit of observation is the filing — not an individual proxy vote — though a single filing may contain thousands of ballot-level vote entries.

Who is required to file Form N-PX?

Registered management investment companies (mutual funds, ETFs, closed-end funds) have been required to file since 2003. Beginning with the reporting period ending June 30, 2024, institutional investment managers subject to Section 13(f) of the Exchange Act and business development companies must also file.

How often are records added to the dataset?

Form N-PX is an annual filing covering the twelve-month period ending June 30, with a filing deadline of August 31. Most new filings arrive in July and August each year. Amendments (N-PX/A) may be filed at any time and appear as separate records.

What time period does the dataset cover?

The earliest records date from August 2003, when the first N-PX filings were due. The dataset is updated continuously and extends to the present.

What file formats are contained in each record?

Structured-era records (reporting periods ending June 30, 2024, onward) contain XML documents and XHTML renderings alongside a JSON metadata file. Pre-2024 records may contain HTML, plain text, or PDF documents. The metadata.json file and folder structure are consistent across all eras.

How does this dataset differ from Form 13F or Form N-PORT data?

Form 13F and Form N-PORT report securities holdings — what a fund or manager owns — but not how they voted. Form N-PX is the only SEC filing that captures security-by-security, proposal-by-proposal proxy voting records at the individual fund level. The three datasets complement each other and can be joined on fund and security identifiers.