Brazil CNPJ Scraper: Company Search & Lookup

This Brazil CNPJ scraper looks up Brazilian companies by CNPJ, or searches the Receita Federal's company register by state, city, CNAE activity code and legal form, and returns one clean row per company: legal name, status, opening date, legal form, size, activities, address, phones and share capital. It is built for teams that need company records without people's data for lead lists, supplier checks, CRM enrichment and market sizing: partners are returned as a count, never as names, and sole traders are skipped and counted. To try it, leave the two sample CNPJs in the input and click Start: the run takes a few seconds.

• Company-level rows only. No partner names, roles or tax numbers. A sole trader, a MEI or another registration of a natural person is never returned; the run summary says how many were skipped.
• You pay for the companies you get, not for the rows that were read. A search row that fails the active, size or opening-date filter is neither stored nor charged. A looked-up CNPJ that is invalid, not in the register or registered to a person costs nothing.
• No start fee of our own. One event per company record and Apify's standard $0.00005 per run start. A run that looks up one CNPJ costs $0.00205.

A row looks like this (shortened; the full example is under Output):

{
    "cnpj": "33.000.167/0001-01",
    "legalName": "PETROLEO BRASILEIRO S A PETROBRAS",
    "status": "active",
    "openedOn": "1966-09-28",
    "legalNature": { "code": "2038", "text": "Sociedade de Economia Mista" },
    "companySize": "other",
    "mainActivity": { "code": "0600001", "text": "Extração de petróleo e gás natural" },
    "address": { "street": "AVENIDA REPUBLICA DO CHILE", "number": "65", "city": "RIO DE JANEIRO", "state": "RJ" },
    "phones": ["+552121660000"],
    "shareCapital": 205431960000,
    "partnersCount": 8
}

How to look up a CNPJ or search Brazilian companies

1. To look up companies, paste their CNPJs under CNPJs to look up, one per line, with or without punctuation (33.000.167/0001-01 or 33000167000101). The alphanumeric CNPJ issued since July 2026 (12.ABC.345/01DE-35) works too. Check digits are verified before anything is requested.
2. To search, fill at least one field under Search the register: states, a city (by name or by IBGE code), CNAE activity codes or legal forms. Then narrow the result with Active companies only, Company sizes or Opened on or after, and set Maximum search results.
3. You can do both in one run. Lookups run first; a company found by both is stored once.
4. Click Start. Rows are saved while the run works, so a run you stop keeps what it has.
5. Open the Output tab and export the table as JSON, CSV or Excel, or read it through the API.

Pricing

You pay per company record: a dataset row with "found": true.

Apify plan	Per 1,000 company records	Per record
Free, Bronze	$2.00	$0.002
Silver	$1.75	$0.00175
Gold	$1.50	$0.0015

Plus Apify's standard $0.00005 per run start. Platform usage is included, so there is nothing else to pay.

• A search for 1,000 active software companies in the state of São Paulo (states: ["SP"], cnaeCodes: ["6201501"]) read 2,251 register rows on 2026-10-07, dropped 873 companies that were not active and 378 sole traders, and stored 1,000 records: $2.00 on the Free and Bronze plans, $1.75 on Silver, $1.50 on Gold. The 1,251 rows that were read and dropped cost nothing.
• A list of 200 CNPJs, of which 180 are companies, 12 are sole traders, 5 are not in the register and 3 have a typing mistake: 180 charges, $0.36 on the Free and Bronze plans. The other 20 cost nothing: 17 get a row that says why, and the 3 mistyped ones are named in the run summary.

What a run that stops early costs. A record is charged at the moment its row is saved, and only then. A run that fails, is aborted or is moved to another server has charged exactly the company records ("found": true) in its dataset. When it is restarted or resurrected it continues where it was: a CNPJ that already has a row is not fetched, stored or charged again. At most one search page is fetched a second time.

You can set a maximum charge per run. The Actor then starts no lookup and fetches no page that the limit could not pay for, stops when the limit is reached and says so in its status message.

Input

Field	What it does
CNPJs to look up (cnpjs)	CNPJs with or without punctuation, numeric or alphanumeric. Up to 5,000 per run. Repeats are looked up once. An entry whose check digits do not fit is skipped without a request.
States (states)	Two-letter state codes (UF). Runs at the source.
City (city)	One municipality by name (São Paulo, campinas) or 7-digit IBGE code (3550308). A name that exists in several states needs the state: Bom Jesus, PI. Runs at the source.
CNAE activity codes (cnaeCodes)	7-digit CNAE subclass codes, 6201501 or 6201-5/01. Matches the main activity. Runs at the source.
Also match secondary activities (includeSecondaryCnae)	Default off. When on, a company matches when the code is its main or one of its secondary activities.
Legal forms (legalNatures)	4-digit legal-nature codes, for example 2062 (Sociedade Empresária Limitada) or 2054 (Sociedade Anônima Fechada). Runs at the source. Codes of natural persons are refused.
Active companies only (activeOnly)	Default on. Applied by the Actor to the rows it fetched. Search only.
Company sizes (companySizes)	micro, small, other. Empty keeps every size. Applied by the Actor. Search only.
Opened on or after (openedAfter)	A date, YYYY-MM-DD. Applied by the Actor. Search only; use it with a narrow search.
Maximum search results (maxResults)	Default 100, at most 100,000. The search stops when this many companies are stored. Lookups are not counted.

A search needs at least one of states, city, cnaeCodes or legalNatures: those are the only filters the register source has. Input that cannot be used (an unknown state, an ambiguous city, a CNAE code with fewer than 7 digits, nothing to do) fails the run before any request, with one sentence that says what to enter instead.

Look up three companies:

{
    "cnpjs": ["33.000.167/0001-01", "00.000.000/0001-91", "60.746.948/0001-12"]
}

Active small and larger software companies in Campinas, opened since 2024:

{
    "city": "Campinas",
    "cnaeCodes": ["6201501", "6202300"],
    "activeOnly": true,
    "companySizes": ["small", "other"],
    "openedAfter": "2024-01-01",
    "maxResults": 200
}

Limited companies and corporations in two states:

{
    "states": ["SC", "PR"],
    "legalNatures": ["2062", "2054"],
    "cnaeCodes": ["4711302"],
    "includeSecondaryCnae": true,
    "maxResults": 5000
}

Output

One row per CNPJ. cnpjDigits is unique within a run. A head office and each of its branches have their own CNPJ and their own row (isHeadOffice tells them apart). Every field is always present; null means the register does not say. Lookups come out in the order of the input, search results in the order the source returns them (not by name or date).

[
    {
        "cnpj": "33.000.167/0001-01",
        "cnpjDigits": "33000167000101",
        "found": true,
        "outcome": "found",
        "note": null,
        "legalName": "PETROLEO BRASILEIRO S A PETROBRAS",
        "tradeName": "PETROBRAS - EDISE",
        "status": "active",
        "statusRaw": "ATIVA",
        "statusDate": "2005-11-03",
        "statusReason": null,
        "specialStatus": null,
        "openedOn": "1966-09-28",
        "legalNature": { "code": "2038", "text": "Sociedade de Economia Mista" },
        "companySize": "other",
        "shareCapital": 205431960000,
        "mainActivity": { "code": "0600001", "text": "Extração de petróleo e gás natural" },
        "secondaryActivities": [
            { "code": "1921700", "text": "Fabricação de produtos do refino de petróleo" },
            { "code": "3520401", "text": "Produção de gás; processamento de gás natural" }
        ],
        "address": {
            "street": "AVENIDA REPUBLICA DO CHILE",
            "number": "65",
            "complement": null,
            "district": "CENTRO",
            "city": "RIO DE JANEIRO",
            "cityCode": "3304557",
            "state": "RJ",
            "postalCode": "20031170"
        },
        "phones": ["+552121660000"],
        "email": null,
        "isHeadOffice": true,
        "simplesOptant": null,
        "taxRegimes": [
            { "year": 2023, "regime": "LUCRO REAL" },
            { "year": 2024, "regime": "LUCRO REAL" }
        ],
        "partnersCount": 8,
        "registryDataDate": "2026-09",
        "source": "minhareceita.org",
        "sourceUrl": "https://minhareceita.org/33000167000101",
        "scrapedAt": "2026-10-07T21:36:12.420Z"
    },
    {
        "cnpj": "00.000.000/0001-91",
        "cnpjDigits": "00000000000191",
        "found": true,
        "outcome": "found",
        "note": null,
        "legalName": "BANCO DO BRASIL SA",
        "tradeName": "DIRECAO GERAL",
        "status": "active",
        "statusRaw": "ATIVA",
        "statusDate": "2005-11-03",
        "statusReason": null,
        "specialStatus": null,
        "openedOn": "1966-08-01",
        "legalNature": { "code": "2038", "text": "Sociedade de Economia Mista" },
        "companySize": "other",
        "shareCapital": 120000000000,
        "mainActivity": { "code": "6422100", "text": "Bancos múltiplos, com carteira comercial" },
        "secondaryActivities": [
            { "code": "6499999", "text": "Outras atividades de serviços financeiros não especificadas anteriormente" }
        ],
        "address": {
            "street": "QUADRA SAUN QUADRA 5 BLOCO B TORRE I, II, III",
            "number": "SN",
            "complement": "ANDAR T I SL S101 A S1602 T II SL C101 A C1602 TIII SL N101 A N1602",
            "district": "ASA NORTE",
            "city": "BRASILIA",
            "cityCode": "5300108",
            "state": "DF",
            "postalCode": "70040912"
        },
        "phones": ["+556134939002"],
        "email": null,
        "isHeadOffice": true,
        "simplesOptant": false,
        "taxRegimes": [
            { "year": 2023, "regime": "LUCRO REAL" },
            { "year": 2024, "regime": "LUCRO REAL" }
        ],
        "partnersCount": 41,
        "registryDataDate": "2026-09",
        "source": "minhareceita.org",
        "sourceUrl": "https://minhareceita.org/00000000000191",
        "scrapedAt": "2026-10-07T21:36:12.745Z"
    },
    {
        "cnpj": "12.ABC.345/01DE-35",
        "cnpjDigits": "12ABC34501DE35",
        "found": false,
        "outcome": "not_found",
        "note": "The register has no company with this CNPJ.",
        "legalName": null,
        "tradeName": null,
        "status": null,
        "statusRaw": null,
        "statusDate": null,
        "statusReason": null,
        "specialStatus": null,
        "openedOn": null,
        "legalNature": null,
        "companySize": null,
        "shareCapital": null,
        "mainActivity": null,
        "secondaryActivities": [],
        "address": {
            "street": null,
            "number": null,
            "complement": null,
            "district": null,
            "city": null,
            "cityCode": null,
            "state": null,
            "postalCode": null
        },
        "phones": [],
        "email": null,
        "isHeadOffice": null,
        "simplesOptant": null,
        "taxRegimes": [],
        "partnersCount": null,
        "registryDataDate": "2026-09",
        "source": "minhareceita.org",
        "sourceUrl": null,
        "scrapedAt": "2026-10-07T21:37:02.118Z"
    }
]

The first two rows are the output of the sample input on 2026-10-07, with the lists of secondary activities and tax regimes shortened. The third is what a valid CNPJ that is not in the register gives: a row that says so, and no charge.

What outcome means

outcome	found	Charged	Meaning
found	true	yes	A company record. Every search result is one.
not_found	false	no	The register has no such CNPJ.
natural_person_skipped	false	no	The CNPJ belongs to a sole trader, a MEI or another registration of a person. The row carries the number and the reason, nothing else.

The run summary

The key-value store record RUN_SUMMARY holds the same facts for programs: recordsStored and chargedEvents, scanned, filteredOut, soleTradersSkipped, scanCapReached, the reason the run stopped (stopReason), and for lookups the invalid entries and the CNPJs that could not be looked up, each with its reason.

Where each filter runs

The register source can filter by state, municipality, CNAE code and legal form, and by nothing else. So:

Filter	Runs	What that means
states, city, cnaeCodes, legalNatures	at the source	Only matching rows are read.
activeOnly, companySizes, openedAfter	in the Actor	Matching rows are read page by page, and the ones that fail are dropped. Dropped rows are not charged.

About half of the register is closed, suspended or unfit companies and close to a fifth is sole traders: of 1,000 rows sampled on 2026-10-07, 421 were active companies. With activeOnly on, the Actor reads between 2 and 2.5 rows for every record it stores.

Limits

• People's data is left out, on purpose. The register's partner list (names, roles, masked CPFs, entry dates, age bands) is not returned; partnersCount is the only thing taken from it. If you need partner names for KYC or to address a person, this Actor is the wrong tool.
• No search by a partner's CPF. That is a search for a person.
• Sole traders, MEI and other registrations of natural persons are never returned (legal nature 2135, the 4000 to 4999 block, any MEI, and any legal name that carries a CPF). Their legal name is a person's name. The same holds for two registrations that are no legal entity: consortia of rural employers (2283, registered as "a person's name and others") and notary offices (3034, a public office delegated to a person). All of them are skipped and counted in soleTradersSkipped. A company named after its owner, or a company phone that is somebody's mobile, can still point at a person: that is how the register is.
• What the register says about an owner is left out. statusReason and specialStatus are null when the register's text describes a person and not the company: the estate or death of an owner, or an owner who is incapacitated or under interdiction.
• No search by company name. The source has none.
• email is null in practice. The current register extract carries no email addresses (0 of 1,000 sampled rows on 2026-10-07). The field is there for the day it does.
• Status, size and opening date are not filtered at the source. They run in the Actor, as described above. openedAfter is only practical on a narrow search (one city or one CNAE code): the source does not sort by date, so on a broad search most rows are read and dropped.
• The scan cap. A search may read 20,000 rows plus 200 for every company it has stored, and at most 300,000 rows in one run. A search whose filters keep fewer than about 1 row in 200 therefore ends early, with scanCapReached: true and whatever it found. Narrow it with a state, a city or a CNAE code.
• Not real time. The source follows the Receita Federal's monthly extract. registryDataDate on every row says which month (2026-09 when this was written). A company opened or closed after that is not reflected yet.
• One database, no service level. Records come from minhareceita.org, a volunteer-run mirror of the register, asked one request at a time with a pause. When it does not answer a lookup (timeouts, server errors or "too many requests" that outlast the retries), the lookup is tried through brasilapi.com.br, which serves the same data, at one request a second and for at most 200 CNPJs a run; source says which path answered. The second path is never used after "not found" or after a refusal, and never for search.
• No lookups through publica.cnpj.ws. Its terms forbid passing on the data of its API commercially.
• No lookups through open.cnpja.com. Its terms could not be read with a plain request, so it is not used.
• A refusal is final. If the source answers 403 or shows a bot check, the run stops and says so. The Actor uses no proxy and no browser and does not try to get around a refusal. When the source is down, the run keeps what it stored and can be resurrected to continue. It is marked failed whenever a refusal or an outage stopped it before the work was done, whatever it stored first, and when more of its CNPJs went without an answer than were answered.
• A code that does not exist gets no answer. The source answers a search for a real CNAE or legal-nature code in well under a second, and does not answer at all when the code does not exist and no state or city narrows the search (measured 2026-10-08). The run then waits for its deadline, fails and says to check the codes. A code made of zeros is refused before any request.
• Paged search is marked "in test phase" by the source. Its format may change. The Actor checks every page before reading it and fails with a clear message rather than returning wrong rows.
• Fields that are not returned: the partner graph, any risk score, derived company age, fax, the dates of entering or leaving Simples and MEI, foreign-address fields and the public-body field.
• Sizes and speed. 5,000 CNPJs and 100,000 search results per run. Lookups are sent one at a time with a quarter-second pause, about 150 a minute when the source answers as fast as it did on 2026-10-07 (0.15 s). A search reads about 1,000 rows every 6 seconds.

FAQ

Is it legal to scrape Brazilian company data?

The CNPJ register is public: the Receita Federal publishes it as open data every month. This Actor reads that data through open, keyless endpoints without logging in, and leaves out the parts that describe people. One of the two hosts, brasilapi.com.br, asks in its terms not to be used by automated tools; the Actor uses it only as the slow fallback described above. You are responsible for how you use the data, including Brazil's data protection law (LGPD) when a record can be linked to a person. This is not legal advice.

Why is a CNPJ "not found" when the company exists?

The data is a monthly extract. A company registered after the month in registryDataDate is not in it yet. Also check the number: the Actor only looks up CNPJs whose check digits are right.

Why did my search return fewer companies than maxResults?

Either the register has no more rows for that search (search.reachedLastPage in the run summary), or the filters dropped most rows and the scan cap ended the run (scanCapReached). filteredOut and soleTradersSkipped show how many rows were read and dropped.

Why is there no row for one of my CNPJs?

Entries that are not a valid CNPJ get no row and no request; they are listed in RUN_SUMMARY under lookup.invalid with the reason. A CNPJ that no source could answer is listed under lookup.failed and has no row either, so that a later run can try it again.

Can I get the partners or the owner of a company?

No. See Limits: the Actor returns how many partners the register lists and nothing about who they are.

Can I search by city name?

Yes. Accents and capitals do not matter. If the name exists in more than one state, the run stops before doing anything and lists the candidates; add the state (Bom Jesus, PI) or use the IBGE code.

How long does a run take?

One lookup takes under half a second, pause included, so 1,000 CNPJs take about 7 minutes. The search for 1,000 active companies in the pricing example took about 20 seconds on 2026-10-07. Both depend on how fast the source answers.

Where do I report a problem?

On the Issues tab of this Actor. Include the run ID.

More Actors from this developer

Company registers:

• North Data Scraper: German & European Companies
• Handelsregister Scraper: German Company Register
• US Business Entity Search & New Business Filings
• European Company Registry Search

Contacts and lists:

• Website Contact Scraper
• Email Validator & List Cleaner

Jobs:

• Company Jobs Search
• ATS Job Postings: Workday, Greenhouse, Lever & Ashby
• Workday Jobs Scraper
• Greenhouse Jobs Scraper
• Lever Jobs Scraper
• Ashby Jobs Scraper