The Form NSAR-A Files Dataset is the complete EDGAR collection of semi-annual reports filed by registered open-end and closed-end management investment companies under Section 30 of the Investment Company Act of 1940. Each record is one EDGAR submission of a Form NSAR-A (original mid-fiscal-year report) or Form NSAR-A/A (amendment), identified by its 18-character SEC accession number and packaged as a flat folder containing a metadata.json header alongside the line-coded answer.fil questionnaire body and any accompanying exhibits. The filer is the registered fund — the trust or corporation — not its adviser, distributor, or service providers. Coverage starts on January 1, 1994 with the earliest EDGAR submissions and continues to the present, with original filings concentrated through June 1, 2018 (when the SEC's investment company reporting modernization rulemaking retired Form N-SAR in favor of Form N-CEN and Form N-PORT) and a residual tail of late originals and NSAR-A/A amendments thereafter. The dataset is distributed as ZIP containers carrying TXT, JSON, HTML, XFD, FRM, FIL, and PDF documents.
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:
The dataset captures every Form NSAR-A and Form NSAR-A/A submission accepted by EDGAR from January 1994 to the present. Form N-SAR was the SEC's standardized questionnaire-style report for registered management investment companies and small business investment companies; the "-A" suffix designates the mid-fiscal-year (six-month interim) installment, with the year-end installment filed on Form NSAR-B. Registrants responded to a fixed schedule of numbered questions ("items") using binary flags, single-letter codes, numeric figures, and short uppercase text strings, producing a highly structured machine-readable record of organizational identity, service-provider relationships, capital structure, sales and redemption activity, portfolio composition and transactions, expenses and fee arrangements, monthly average net assets, and compliance attestations.
The N-SAR body was always filed as a flat ASCII text document wrapped in the EDGAR SGML document envelope. The form was never converted to HTML, never wrapped in XBRL, and never adopted inline XBRL, because the fixed-column item structure already supplied the machine-readable structure those formats provide elsewhere on EDGAR. Supplementary exhibits — auditor attestations, internal-control reports under Rule 17f-2, opinion letters, and similar attachments — could accompany the primary answer file as TXT, HTM/HTML, PDF, XFD, or FRM documents. The questionnaire schedule remained substantially stable from EDGAR's adoption in the early 1990s through the form's retirement, giving the dataset unusually consistent record anatomy across roughly a quarter-century.
A single record in the Form NSAR-A Files dataset is one complete EDGAR submission, physically organized as an accession-named folder (dashes stripped) that holds a metadata.json describing the submission together with the document files that EDGAR received for it. Logically, the record captures one semi-annual report by a registered investment management company under Section 30(b)(1) of the Investment Company Act of 1940 covering the first six months of the registrant's fiscal year — or, in the case of NSAR-A/A, a subsequent amendment to such a report.
Each accession folder bundles two layers of content in a flat directory layout (no nested subfolders):
metadata.json derived from the EDGAR submission header, providing canonical filing identifiers, registrant identity, period of report, filer/series/class identifiers, and an enumeration of every document file in the submission.answer.fil — the SGML-wrapped, line-coded N-SAR questionnaire body — accompanied, when present, by exhibit files (TXT, HTM/HTML, PDF, XFD, FRM) and by the rolled-up complete-submission .txt produced by EDGAR.metadata.json structuremetadata.json is the structured spine of every record. Its meaningful fields are:
formType — "NSAR-A" for an original semi-annual report or "NSAR-A/A" for an amendment.accessionNo — the EDGAR accession number in its canonical dashed form, e.g. "0001140361-18-045903".filedAt — ISO-8601 acceptance timestamp from EDGAR, including the Eastern-time offset.periodOfReport — the report's period-end date in YYYY-MM-DD form, corresponding to the six-month interim cut-off.description — short human label such as "Form NSAR-A - Semi-annual report for management companies".linkToFilingDetails — direct URL to the primary submission document on EDGAR.linkToTxt — URL to the rolled-up complete-submission .txt on EDGAR.linkToHtml — URL to the EDGAR filing-index page (-index.htm).linkToXbrl — present in the shared schema but unused for N-SAR filings.id — internal 32-character hexadecimal identifier for the record.documentFormatFiles — array of every document in the EDGAR submission. Each element carries sequence (submission ordinal as a string, blank for the rolled-up complete-submission text), size (uncompressed byte count as a string), documentUrl, type (EDGAR document-type label such as "NSAR-A"), and description.dataFiles — array of supplementary structured data files; typically empty for N-SAR, since the questionnaire is delivered as a single coded text body rather than as separate data files.entities — array of EDGAR entity records associated with the submission. Each entity carries companyName with a role suffix (e.g. "… (Filer)"), zero-padded ten-digit cik, type (form type recorded against the entity), act (governing statute, "40" for the Investment Company Act of 1940), fileNo (the SEC file number, generally in the 811- series for registered investment companies), filmNo (EDGAR film number), irsNo (employer identification number), and fiscalYearEnd in MMDD form.seriesAndClassesContractsInformation — array of investment-company series and class-contract identifiers (CIK, series S… identifier, class-contract C… identifier). Populated when the filer is a series trust; empty for stand-alone funds and for filings predating the SEC's series/class identifier framework.answer.fil — the line-coded NSAR-A bodyThe primary document is a flat ASCII text file conventionally named answer.fil, wrapped in the EDGAR SGML document envelope:
1
<DOCUMENT>
2
<TYPE>NSAR-A
3
<SEQUENCE>1
4
<FILENAME>answer.fil
5
<DESCRIPTION>NSAR-A
6
<TEXT>
7
<PAGE> PAGE 1
8
... line-coded answers ...
9
<PAGE> PAGE 2
10
...
11
SIGNATURE <name>
12
TITLE <role>
13
</TEXT>
14
</DOCUMENT>
Inside the <TEXT> block, each non-pagination line is a fixed-column response to one N-SAR item. The general pattern is a three-digit item number, an optional sub-item letter, an optional numeric sub-index for repeating rows or sub-fields, and a value. For example, an adviser name on row 2 of item 8 sub-item A appears as 008 A000002 followed by the value in the response column; flag responses appear as Y or N; numeric responses appear as right-justified integers or decimals; short text responses appear in uppercase. Repeating items emit successive lines sharing the same item/sub-item prefix and incrementing the trailing ordinal. <PAGE> markers and the literal PAGE n text are pagination artifacts and carry no semantic content.
Form N-SAR is organized as a numbered questionnaire running from item 1 through item 133, with sub-items addressed by letters (and, where needed, numeric sub-indices). The recurring blocks inside an NSAR-A answer body are:
SIGNATURE <name> and TITLE <role>, precede the closing </TEXT> and </DOCUMENT> tags.Items that do not apply to a given filer (for example, item ranges addressing 12b-1 plan attributes for a fund without a 12b-1 plan, money-market-only items for a non-money-market fund, or per-series schedules for a stand-alone trust) are simply absent from the answer body rather than emitted with empty values. Consumers should treat missing item lines as "not applicable" rather than as zero or false.
The accession folder always includes metadata.json and the primary .fil answer document. Depending on the filing, it may additionally include:
.txt that wraps every document of the submission inside the EDGAR <SEC-DOCUMENT> header — enumerated in documentFormatFiles with a blank sequence..txt) containing auditor opinions, internal-control attestations under Rule 17f-2, fidelity-bond detail, or other narrative attachments..htm, .html) carrying the same exhibit roles when the filer attached HTML rather than plain text..pdf) for scanned or rendered documents such as signed accountant letters..xfd files (EDGAR's XFDL form-data format) used by certain N-SAR submissions for structured form-data attachments..frm files representing EDGAR FRM-format exhibits.The file-types found in the dataset are TXT, JSON, HTML, XFD, FRM, FIL, and PDF. The overwhelming majority of NSAR-A records contain only metadata.json plus answer.fil; richer exhibit assemblages appear in a minority of filings and sit flat inside the accession folder alongside the primary document.
Image attachments (GIF, JPG, and similar binary image formats) that may have accompanied an EDGAR submission are not carried in the dataset. The dataset is otherwise the verbatim, decompressed copy of the EDGAR submission documents. EDGAR-side correspondence, the filer's working files, and any document not submitted as part of the original EDGAR package are likewise outside the record. There is no per-share-class or per-holding breakout beyond what the questionnaire itself encodes; portfolio detail at the individual security level lives in separate reporting regimes (Form N-Q, later Form N-PORT) and is not present here.
Amendments are emitted as independent records under their own accession numbers with formType set to "NSAR-A/A". The internal structure mirrors the original NSAR-A: a metadata.json plus an answer.fil carrying the amended line-coded answer set and any re-attached exhibits. The amendment indicator inside the answer body's header block is set accordingly. Because each amendment carries its own accession number and metadata.json, the link between an amendment and the filing it amends is not encoded as an explicit foreign key inside the record; it is reconstructed by matching the registrant CIK (or series identifier) and the periodOfReport. Successive amendments to the same period appear as separate records sharing CIK and period-end but differing in accessionNo, filedAt, and formType. Point-in-time reconstructions of a registrant's most recently amended semi-annual report require sorting same-period records by filedAt and taking the latest.
The body file format is stable across the dataset's span: a flat ASCII document inside an SGML envelope, line-coded by item number, signed off in uppercase. What evolved was the population of supplementary exhibit formats — early-era filings tend to ship as answer.fil plus, occasionally, plain-text exhibits, while filings from the 2000s and 2010s more often include HTML, PDF, XFD, and FRM exhibits for auditor attestations and internal-control reports. The series/class identifier system (S… for series, C… for class contracts) became meaningful only after the SEC introduced the framework in the mid-2000s; earlier filings carry an empty seriesAndClassesContractsInformation array. The discontinuation of the form after June 1, 2018 produces a sharp drop-off in record volume after mid-2018, with the residual tail comprising late originals and amendments.
<PAGE> line as a tuple of (item number, sub-item letter, numeric sub-index, value) rather than as free text. The <PAGE> markers and accompanying PAGE n strings are pagination artifacts of the SGML envelope.entities[].companyName carries a parenthetical role suffix (e.g. "(Filer)") that distinguishes the registrant from other associated entities recorded by EDGAR; downstream joins on company name should strip this suffix.metadata.json is the trust as a whole, while per-series identifiers are carried in seriesAndClassesContractsInformation and the per-series financial schedules sit at the tail of the answer body. Joining a specific series' financial detail therefore requires correlating the series S… identifier in metadata with the series-numbered items inside answer.fil.The filer is the registered investment management company itself (the trust or corporation), not its adviser, distributor, or service providers. The CIK on the submission belongs to the fund registrant.
Two filer classes use NSAR-A:
Both filer classes use the same form, answering conditional items based on type.
Entities outside the NSAR-A population:
Authority comes from Section 30 of the Investment Company Act of 1940 (15 U.S.C. 80a-29), which empowers the SEC to require periodic reports from registered investment companies. The implementing rules during the N-SAR era were:
N-SAR was a structured, fixed-field data collection covering portfolio composition (broad categories), fees, sales loads, distribution arrangements, service providers, board composition, expense ratios, and operational metrics — not narrative disclosure.
NSAR-A is schedule-driven, not event-driven. The trigger is the close of the first six months of the registrant's fiscal year (the mid-year semi-annual period). There is no materiality test or transaction threshold.
The deadline is 60 days after the end of the reporting period. For a December 31 fiscal year, the mid-year period ends June 30 and NSAR-A is due around August 29. NSAR-B (the annual N-SAR covering the full fiscal year, including audit-related items) is also due 60 days after fiscal year end.
Because deadlines track each registrant's fiscal year, NSAR-A filings appear throughout the calendar year, clustering after common fiscal year-ends (March, June, October, December).
Form N-SAR was adopted in 1985, replacing predecessor Forms N-1R and N-1Q. EDGAR submissions begin in the mid-1990s; the dataset's earliest records date to January 1994. Pre-EDGAR paper filings sit in SEC paper archives outside this dataset.
The form was rescinded effective June 1, 2018, under the SEC's investment company reporting modernization rulemaking (Investment Company Act Release No. 32314, adopted October 13, 2016). After that date, registrants no longer file new NSAR-A or NSAR-B reports. N-SAR's data collection was replaced by:
The dataset is therefore historically bounded: original NSAR-A filings run from 1994 through mid-2018, with any later submissions being amendments of pre-2018 periods.
Form NSAR-A/A is the amendment variant. It is filed by the same registrant to correct, restate, or supplement a previously filed NSAR-A covering the same six-month period. Amendments may be triggered by:
Each NSAR-A/A is a separate EDGAR submission and a separate record in this dataset. There is no statutory cap on how late an amendment may be filed, so NSAR-A/A submissions can post-date the June 1, 2018 rescission when they correct pre-rescission periods.
Form NSAR-A sits within a tightly clustered family of investment-company filings. The sharpest comparisons are with its N-SAR siblings, the post-2018 successor forms (N-CEN, N-PORT), the shareholder-report regime (N-CSR/N-CSRS, N-Q), and the registration-statement family (N-1A/N-2/N-3). The differences below are organized around timing, filer population, content scope, and format.
The single closest comparison. Same filer population, same statutory basis (Section 30 of the 1940 Act), same line-coded answer.fil format, and largely the same line-item universe.
Key differences:
Same statute and cadence, but a disjoint filer population: UITs rather than management investment companies. Because UITs lack an active adviser and managed portfolio, NSAR-U omits or restructures items on portfolio turnover, adviser compensation, and management-company governance. NSAR-A contains no UIT data, and NSAR-U contains no management-company data — treat them as non-overlapping by filer type.
Same content fields as NSAR-A/NSAR-B respectively, but filed for stub periods triggered by fiscal-year changes. The reporting interval is non-standard and the filings appear sporadically. For panel analysis, NSAR-AT observations must be reconciled to the underlying fiscal-year change; mixing them into a regular semi-annual series distorts period-over-period comparisons. The same caveats apply to NSAR-BT filings.
The direct successor to NSAR-B on the operational/census axis. Key contrasts with NSAR-A:
answer.fil.Clarifies what NSAR-A is not. N-PORT is filed monthly in XML and reports position-level holdings, derivatives exposure, liquidity classifications, and risk metrics. NSAR-A contains only aggregate operational metrics (totals, ratios, turnover rates) — no security-by-security holdings. Pre-2018 holdings cannot be reconstructed from NSAR-A; researchers must use N-CSR/N-CSRS or the former N-Q.
N-CSRS shares NSAR-A's semi-annual cadence and filer base, but the documents differ in audience and content:
They are complements for the same period: N-CSRS provides narrative and holdings, NSAR-A provides structured operating metrics. Neither substitutes for the other.
Rescinded like NSAR-A, but content was holdings-focused, filling the off-cycle gap between N-CSR and N-CSRS. Its successor is N-PORT, not N-CEN. The comparison highlights how the legacy regime split disclosure across narrow forms (NSAR-A/B for operations, N-CSR/N-CSRS for shareholder reports, N-Q for interim holdings) — a structure consolidated post-2018 into N-CEN plus N-PORT.
Same filer universe (open-end funds, closed-end funds, and separate accounts organized as management companies), but a different disclosure stage. Registration statements are forward-looking prospectus documents describing objectives, strategies, risks, fees, and governance. NSAR-A is a backward-looking operational return reporting actual mid-year results. Fee tables in N-1A and expense ratios in NSAR-A are cross-referenceable, but N-1A states what the fund intends; NSAR-A states what it did. The same relationship holds for N-2 (closed-end registration) and N-3 (separate-account registration).
NSAR-A is distinct by the combination of four properties that no other dataset reproduces together:
answer.fil format requiring line-number-aware parsing rather than XML or XBRL.NSAR-B is the closest complement within the legacy regime; N-CEN is the closest successor on the operational axis (annual only); N-PORT and the former N-Q occupy a holdings axis NSAR-A never covered; N-CSR/N-CSRS occupy the shareholder-narrative axis; and N-1A/N-2/N-3 occupy the offering-document axis. For mid-year operational data on management investment companies between 1994 and 2018, no alternative substitutes for NSAR-A.
NSAR-A captures semi-annual operating data for registered open-end and closed-end funds from 1994 through mid-2018, when N-CEN and N-PORT replaced it. The dataset is used by professionals who need a continuous, series-level view of the pre-modernization U.S. fund industry — flows, fees, turnover, NAV, service providers, and compliance attestations.
Used for peer-reviewed work on fee dispersion, scale economies, flow-performance sensitivity, and after-cost returns. Researchers pull Items 028 to 031 (sales, redemptions, exchanges) for flow panels, Items 042 to 048 (management fees, 12b-1, expenses) for fee studies, Items 070 to 072 (purchases, sales, turnover) for trading-cost work, and Items 074 to 076 (NAV, total net assets) for size normalization.
Quants backfill fund-characteristic factors that commercial vendors only carry from later dates. They use Items 074 to 076 for long-run AUM series, Items 028 to 031 for flow-momentum and capacity models, Items 070 to 072 for turnover features, and Items 042 to 048 for net-of-fee performance reconstruction. Output feeds factor pipelines and active-strategy capacity models.
Used to reconstruct registrant operating history for SEC examinations, fee litigation defense, and reorganization due diligence. Items 001 to 008 trace registrant, series, and class identifiers across successions and mergers. Items 010 to 020 evidence adviser, sub-adviser, underwriter, custodian, transfer agent, and auditor history. Item 077 attestations on sales loads, 12b-1 plans, Rule 17a-7, and affiliated brokerage are reviewed alongside the NSAR-A/A amendment population to verify restatements.
Used as primary source material in excessive-fee, market-timing, and late-trading cases when contemporaneous records are unavailable. Items 042 to 048 establish expense-ratio paths for Section 36(b) damages; Items 028 to 031 quantify flow consequences of alleged disclosure failures; Items 070 to 072 support trading-behavior analysis; Items 010 to 020 identify responsible service providers; Items 074 to 076 supply the asset base for fee-damages calculations. Output feeds expert reports and deposition exhibits.
Used as the upstream feed for pre-2018 series-level panels, bridged to N-CEN and N-PORT for 2018-forward continuity. Engineers parse identification (001 to 008), service providers (010 to 020), flows (028 to 031), expenses (042 to 048), portfolio activity (070 to 072), and NAV/assets (074 to 076) into normalized tables keyed by CIK, series ID, and class ID, so downstream consumers query one longitudinal panel without a 2018 gap.
Used in retrospective rule reviews and cost-benefit assessments of the reporting modernization rulemaking. Items 042 to 048 support long-run fee and 12b-1 trend studies; Items 070 to 072 inform turnover distributions; Items 028 to 031 quantify historical flow patterns; compliance-attestation items reveal exemptive-order reliance and affiliated-transaction activity. Feeds sweep exam targeting and economic analyses for new rule proposals.
Used to fill series-level historical fields missing from commercial fund databases. Items 074 to 076 build AUM histories, Items 042 to 048 reconstruct fee schedules and 12b-1 splits, Items 028 to 031 produce organic-growth and redemption metrics, Items 070 to 072 support peer turnover comparisons, and Items 010 to 020 feed service-provider league tables. Output supports board 15(c) fee benchmarking, RFP responses, and long-history industry studies.
Used to give models structured, citation-ready access to pre-2018 fund operating data. Item-numbered fields are indexed so queries about fee history, flow history, turnover, or service-provider succession at the series level resolve to specific filed records, supporting compliance research copilots and literature-review tools that ground answers in NSAR-A facts.
These users share one requirement: a continuous, series-level view of registered fund operations from 1994 to 2018 that N-CEN and N-PORT do not cover. They pull different item ranges — flows in 028 to 031, expenses in 042 to 048, portfolio activity in 070 to 072, assets in 074 to 076, and service providers in 010 to 020 — but all depend on NSAR-A to make the pre-modernization era analytically usable.
The NSAR-A corpus supports a small set of concrete workflows that depend on its line-coded semi-annual operating data for registered management investment companies from 1994 through mid-2018.
Parse Items 028 to 031 from answer.fil (monthly share sales, redemptions, repurchases, and exchanges) and join to metadata.json keys (entities[].cik, seriesAndClassesContractsInformation[], periodOfReport) to assemble a series-level six-month flow panel. Combined with the NSAR-B annual counterpart, this produces a continuous flow history that bridges directly into N-CEN/N-PORT for 2018-forward analysis. Output feeds flow-performance regressions, capacity models, and organic-growth metrics in commercial fund databases.
Extract the management-fee breakpoint ladder from Items 054 to 061, 12b-1 plan attributes from Items 042 to 048, and monthly average net assets from Item 075 to rebuild a fund's effective fee path across every semi-annual period. Item 076 (period-end NAV per share) and Item 074 (total net assets) supply the asset base for damages calculations. Output drops into expert reports and deposition exhibits where contemporaneous prospectus and shareholder-report data alone are insufficient.
Iterate Items 007 to 018 (advisers, sub-advisers, principal underwriters, custodians, sub-custodians, transfer agents, shareholder-servicing agents, administrators, independent public accountants) across the full corpus to produce per-period rosters keyed by CIK and series identifier. Sorting same-registrant filings by periodOfReport and filedAt exposes adviser changes, custody transitions, and auditor rotations. The same workflow supports 15(c) board materials, RFP competitive intelligence, and pre-examination service-provider history packets.
Pull the Item 070 portfolio-transaction matrix (purchases and sales by instrument class) together with Items 071 and 072 (aggregate portfolio activity and turnover) to build mid-year turnover distributions by Lipper-style peer set. The asset-category percentages in Items 062 to 069 segment funds by mandate without relying on third-party classifications. Output supports trading-cost research, peer-turnover comparisons in 15(c) packages, and capacity ceilings for active-strategy backtests.
Group records sharing entities[].cik and periodOfReport, then sort by formType (NSAR-A versus NSAR-A/A) and filedAt to reconstruct the latest amended view of any six-month period. Diffing the line-coded answers between an original and its amendments isolates which items (fees, flows, attestations) were restated, supporting compliance restatement reviews and forensic accounting workups where the timing of the correction matters.
For series companies, correlate the S… series and C… class-contract identifiers in seriesAndClassesContractsInformation with the per-series schedules at the tail of the answer body (Items 089 to 133) to attribute assets, sales, redemptions, and NAV to the correct series rather than the trust as a whole. This is the only practical bridge from trust-level CIK to series-level operations in the pre-2018 record and is required to join NSAR-A history cleanly to N-CEN and N-PORT, which are natively series-keyed.
Index each (accessionNo, item, sub-item, sub-index, value) tuple from answer.fil alongside metadata.json identifiers so retrieval systems answer fee-history, flow-history, turnover, and service-provider succession questions with direct citations to filed records. The deterministic item-number schema lets LLM tools cite specific lines (for example "Item 075 column 6, accession 0001140361-18-045903") rather than paraphrasing narrative disclosures.
The Form NSAR-A Files dataset is available through three access methods: a JSON index endpoint for metadata and container discovery, a full archive download, and direct per-container downloads.
Dataset Index JSON API: https://api.sec-api.io/datasets/form-nsara-files.json
This endpoint returns dataset-level metadata along with the list of monthly container files. Each container entry includes its key, size in bytes, record count, updated timestamp, and download URL. Use this endpoint to monitor which containers changed in the most recent refresh run and selectively download only the updated archives. The endpoint does not require an API key.
Example response:
1
{
2
"datasetId": "1f13365b-9ae0-68f8-9dbd-f708173eb663",
3
"datasetDownloadUrl": "https://api.sec-api.io/datasets/form-nsara-files.zip",
4
"name": "Form NSAR-A Files Dataset",
5
"updatedAt": "2026-05-18T03:12:44.000Z",
6
"earliestSampleDate": "1994-01-01",
7
"totalRecords": 227628,
8
"totalSize": 778876619,
9
"formTypes": ["NSAR-A", "NSAR-A/A"],
10
"containerFormat": "ZIP",
11
"fileTypes": ["TXT", "JSON", "HTML", "XFD", "FRM", "FIL", "PDF"],
12
"containers": [
13
{
14
"downloadUrl": "https://api.sec-api.io/datasets/form-nsara-files/2018/2018-06.zip",
15
"key": "2018/2018-06.zip",
16
"size": 4185293,
17
"records": 142,
18
"updatedAt": "2026-05-18T03:12:44.000Z"
19
}
20
]
21
}
Download Entire Dataset: https://api.sec-api.io/datasets/form-nsara-files.zip?token=YOUR_API_KEY
Downloads the complete dataset as a single ZIP archive covering all NSAR-A and NSAR-A/A filings from January 1994 through June 2018 (with sparse records in the second half of 2018). This endpoint requires a valid SEC API key passed as the token query parameter.
Download Single Container: https://api.sec-api.io/datasets/form-nsara-files/2018/2018-06.zip?token=YOUR_API_KEY
Downloads one monthly container ZIP instead of the full archive. Replace the year and month path segment with any container key listed under containers[].key in the dataset index JSON. This endpoint requires a valid SEC API key.
The dataset covers Form NSAR-A (the original semi-annual report for registered management investment companies) and Form NSAR-A/A (amendments to previously filed NSAR-A reports). Both form types are accepted by EDGAR under the same submission schema.
One record is a complete EDGAR submission of a single NSAR-A or NSAR-A/A filing, identified by its 18-character SEC accession number. Each record is a flat folder containing a metadata.json header and the submission documents — most commonly the line-coded answer.fil questionnaire body plus any exhibits filed alongside it.
Open-end management investment companies (mutual funds, including ETFs organized as open-end funds) and closed-end management investment companies (including listed closed-end funds and interval funds) filed Form NSAR-A. Unit investment trusts, business development companies, investment advisers, private funds relying on Section 3(c)(1) or 3(c)(7), and unregistered foreign funds are outside the NSAR-A population.
The SEC's investment company reporting modernization rulemaking rescinded Form N-SAR effective June 1, 2018, replacing it with Form N-CEN (annual census) and Form N-PORT (monthly portfolio holdings). Original NSAR-A filings therefore end in mid-2018, but the dataset still receives NSAR-A/A amendments to pre-rescission periods because there is no statutory cap on how late an amendment may be filed.
The dataset's earliest records date to January 1, 1994, when EDGAR submissions for Form N-SAR began. Original NSAR-A filings run continuously through June 1, 2018, with later amendments and a sparse tail of late originals continuing thereafter.
The dataset is distributed as ZIP containers organized by month. Inside each container, every record is a flat accession-named folder containing metadata.json plus the EDGAR submission documents in their original formats — TXT, HTML, XFD, FRM, FIL, and PDF.
Both share the same filer population, statutory basis (Section 30 of the Investment Company Act of 1940), and line-coded answer.fil format. NSAR-A covers the first six months of the registrant's fiscal year; NSAR-B covers the full fiscal year and adds annual-only items such as auditor identification, internal-control matters, fidelity-bond coverage, and certain affiliate-transaction disclosures. The two forms are complements, not substitutes.