SIC vs NAICS: Differences, Crosswalks and Which to Use

Use NAICS for anything current. It is the standard the U.S. federal statistical system has used since 1997, it has six digits instead of four, and it is updated on a schedule. SIC (Standard Industrial Classification) is frozen. It survives mainly in legacy databases, some regulatory filings and long historical series. The two systems cannot be converted one for one, so moving between them takes a crosswalk and some judgment.

This guide covers how the systems differ, why conversion is messy, how to map codes properly, and which one to pick for sales prospecting, market sizing, underwriting or research.

TL;DR

• NAICS replaced SIC as the U.S. federal statistical standard starting in 1997. SIC dates from 1937 and was last revised in 1987, per the Census Bureau.
• Structure differs: SIC has four digits. NAICS has six, and the first five are shared with Canada and Mexico.
• There is no one-to-one conversion. One SIC code can split across several NAICS industries, and several can merge into one. Use a weighted crosswalk when accuracy matters.
• Default to NAICS for current work. Keep SIC only when a source, a filing or a decades-long trend line demands it.
• Store both in your CRM or research database if you can, so you can filter by whichever one a list or a partner uses.

Table of Contents

• What SIC and NAICS actually are
• SIC vs NAICS side by side
• Why the grouping logic matters
• Reading the digits
• Why you cannot convert SIC to NAICS cleanly
• How to map with a weighted crosswalk
• Where SIC still shows up
• Which one should you use?
• Our view: codes are a filter, not a verdict
• Finding the code for a specific company
• Putting codes to work in prospecting and research
• Common mistakes
• Doing this at scale with Milda
• FAQ
• Sources

What SIC and NAICS actually are

Both systems answer the same question: what is this business’s primary activity? A code lets you group firms, count them, compare them and filter them. The Washington Department of Revenue describes both as identifying a firm’s primary business activity, and it still publishes guidance on each.

SIC was established in 1937 and last updated in 1987. It was built for an economy dominated by manufacturing, so it is rich in factory and farm categories and thin on services and technology.

NAICS (North American Industry Classification System) was developed jointly by the statistical agencies of the United States, Canada and Mexico so that industry statistics would be comparable across the three countries. It is the standard federal statistical agencies use to classify business establishments, according to the Census Bureau. Updates arrive every five years, and NAICS 2022 is the most recent major revision as of October 2026.

One caveat from the Federal Register notice: NAICS is designed and maintained solely for statistical purposes. Regulators and insurers may use it for other things, but the code itself does not define eligibility or legal status.

SIC vs NAICS side by side

The industry counts are in the same range, which surprises people. The gain from NAICS is not the number of categories but where they sit. NAICS spends its detail on information, professional services, health care and other service industries that SIC barely distinguished. The division and industry counts above come from the Census Bureau’s NAICS overview. Other sites quote different totals, usually because they count a different NAICS vintage or include sub-industry levels, so check the vintage before you cite a figure.

Why the grouping logic matters

The Office of Management and Budget’s rationale is explained in the NAICS manual. NAICS classifies establishments by the production process they use, a supply-side principle. SIC leaned toward what the establishment sells and who buys it.

In practice that changes where a company lands:

• Two firms making the same product by different processes may sit in different NAICS industries.
• Two firms using the same process to serve different markets may sit together in NAICS but apart in SIC.
• A business model that did not exist in 1987, such as e-commerce or software delivered over the internet, has a natural home in NAICS. In SIC you usually pick the nearest old category.

The process-based approach also ages better. When the internet turned from a specialized business into a generic way of delivering services, NAICS did not need a new classification for every retailer that moved online.

Reading the digits

NAICS nests like a tree. The levels, per the 2022 manual:

1. Sector, 2 digits
2. Subsector, 3 digits
3. Industry group, 4 digits
4. NAICS industry, 5 digits
5. U.S. industry, 6 digits

The first five digits are standardized across the three countries. The sixth digit is reserved for national detail, according to Statistics Canada. A six-digit U.S. code may therefore describe an industry that does not exist, or means something different, in Canada or Mexico. If you are comparing across borders, compare at five digits.

SIC is simpler: a two-digit major group, then a three-digit industry group, then the four-digit industry. The Washington Department of Revenue’s example is a firm with SIC 571, which primarily sells home furnishings. That is useful for a quick sort and too coarse for most modern segmentation.

Why you cannot convert SIC to NAICS cleanly

The question “how do I convert a SIC code to NAICS?” has an honest answer: you can map it, but you cannot convert it. The NAICS manual explains that the systems rest on different conceptual frameworks. A single SIC code may be split into several NAICS industries. Several SIC codes may be merged into one NAICS industry. Precise historical comparison needs weighted concordance tables.

Here are the four relationships you will meet in any crosswalk:

• One to one: the rare clean case. Safe to relabel.
• One to many: a SIC industry splits. You must decide which NAICS code each company belongs to by looking at what it does.
• Many to one: several SIC industries merge. Relabeling is easy, but you lose the old distinction.
• Many to many: both at once. Most of the difficulty lives here.

A worked example of the problem

Suppose you hold a list of 5,000 companies tagged with a SIC code that a crosswalk shows mapping to three NAICS industries. If you assign everyone to the first listed NAICS code, you have systematically misclassified the other two thirds. If you instead look at each firm’s description, products and revenue lines, you can place them. The first approach is fast and wrong. The second is slower and defensible.

How to map with a weighted crosswalk

Researchers bridging the 1997 transition use weighted crosswalks, which distribute values across the new classifications using employment, payroll or establishment counts. The ICPSR weighted crosswalk collection is one such resource, and its summary says it exists to bridge SIC and NAICS codes. Its files come in pairs for each code transition, one bridging forward from old to new and one backward, per the project folder.

A sound workflow looks like this:

1. Decide the goal. Are you relabeling a company list, or rebuilding a time series? They need different methods.
2. For company lists, use the crosswalk to find the candidate NAICS codes, then pick one per company from its actual activity.
3. For time series, use the weighted version so that counts or dollars are split in proportion, not duplicated or dropped.
4. Record the vintage. Write down which NAICS year (1997, 2002, 2007, 2012, 2017 or 2022) you mapped to. Codes change between revisions.
5. Spot check. Pull 50 records, read what the companies do, and confirm the new code makes sense.

Simple unweighted mappings can lead to significant errors in time-series work across the transition. If the analysis will be published or used to allocate money, use the weighted approach.

Mind the dataset’s own conventions

Even among modern systems, mappings are not always tidy. The Bureau of Labor Statistics’ Current Employment Statistics program uses industry codes based on NAICS that can combine multiple NAICS codes to handle sample size limits, as described on its published series page. Read the documentation of the specific dataset before assuming a standard table applies. The CES program itself currently uses NAICS 2022.

Where SIC still shows up

Federal statistics left SIC behind long ago. The BLS Current Employment Statistics program began publishing under NAICS with the release of May 2003 data, and it no longer produces SIC-based data, per its series page. Yet you will still run into four-digit codes:

• Securities filings. Public companies are still assigned SIC codes for their SEC filings through EDGAR, according to this overview of why SIC persists.
• Private business databases and legacy software. Many were built on SIC and never migrated.
• Insurance and underwriting workflows, where old appetite guides are keyed to SIC.
• State agency guidance. The Washington Department of Revenue, for one, still explains both systems side by side.
• Long-run historical analysis. The Library of Congress industry research guide notes that legacy consistency across decades can matter more than modern granularity in some niches.

In short, SIC is a legacy label, not a live standard. Treat it as a field you carry, not the one you build on.

Which one should you use?

The pattern is simple. NAICS is the working language. SIC is for matching what others already wrote down.

Our view: codes are a filter, not a verdict

We build company intelligence, so we see the same mistake from every kind of team. They treat an industry code as a description of the company. It is a bucket. A single code lumps together a ten-person specialist and a conglomerate division, and it rarely tells you what the company is growing into.

Three habits have served our clients well:

• Use codes to narrow, then read the company. A NAICS filter gets you from hundreds of millions of companies to a manageable pool. Products, technology portfolio, ownership and recent activity decide who is actually relevant. We make the same argument about firmographics as only the first filter.
• Expect a company to straddle codes. Modern firms sell software, hardware and services at once. One primary code can hide the activity you care about.
• Keep the original code and add the new one. Overwriting SIC with NAICS destroys information you may need when a partner, regulator or old report speaks the other language.

That is why we treat classification as a layer on top of richer signals: ownership, financials, patents, trade data and event-driven growth signals. The code tells you roughly where a company sits. The rest tells you whether to call.

— Vilius Stanciauskas, Founder and CEO

Finding the code for a specific company

Three routes, in order of reliability:

1. Check the official NAICS search. The Census Bureau hosts the code structure and lets you look up a description by keyword. Pick the code whose definition matches the firm’s largest revenue source, not its marketing language.
2. Look at filings. For public companies, the SIC code on the SEC filing page is the quickest legacy reference. Then map it to NAICS yourself.
3. Use a company database that carries both. If you are classifying hundreds of firms, a platform with both systems on each profile saves the manual mapping. You can see how a profile presents this kind of data on an example company profile.

The rule in all three: classify by primary activity, meaning the activity that generates the most revenue, and write down your reasoning for borderline cases.

Putting codes to work in prospecting and research

Industry codes are most useful when you pair them with other criteria.

Sales and business development. Start with NAICS to define the territory, then layer size, geography, ownership and growth signals. Our sales and GTM research page shows how teams build target lists that way. Where your CRM holds SIC, add a NAICS field instead of replacing it.

Market analysts. NAICS lets you line up your estimates with Census and BLS tables, which makes your numbers checkable. Our market research use case covers mapping a category end to end. For markets that fall between codes, such as climate tech or AI services, codes alone undercount. Search by products and technology instead.

Investors and innovation scouts. Industry codes tell you which bucket a company is in but not which companies are new, growing or tied to a given technology. Scouting for partners usually starts from capability, then uses the code as a sanity check.

Common mistakes

• Assuming the six-digit code is international. Only the first five digits are shared across the three NAICS countries.
• Mixing vintages. A NAICS 2012 code and a NAICS 2022 code can differ. Label the year.
• Treating a crosswalk as a lookup table. It is a menu of candidates.
• Using SIC to find modern service companies. Four digits cannot separate many digital and professional services.
• Overwriting source codes. Keep the original beside the mapped one.

Doing this at scale with Milda

If you are mapping or filtering thousands of companies, the manual route stops working. Milda gives you access to more than 420 million companies and 50 million verified contacts, with firmographics, ownership, financials, patents, trade data and growth signals in one place. You can search through the web app, the browser extension or the API, all part of the Data Platform. See the data coverage to check what is available for your target industries, and the pricing page for Browser Analyst, Strategy Terminal and Data Solutions.

FAQ

Is NAICS the same as SIC?

No. They do the same job, classifying a business by its primary activity, but they differ in age, structure and logic. SIC has four digits and dates from 1937, while NAICS has six digits and replaced it for federal statistics from 1997. NAICS also groups firms by production process instead of by product sold.

Should I use SIC or NAICS?

Use NAICS for current analysis, government work and cross-border comparison. Keep SIC when you must match an SEC filing, an older database or a historical series. If you can, store both codes for each company.

How do I convert a NAICS code to a SIC code?

There is no exact conversion, because one code often maps to several in the other system. Use a published crosswalk, such as the ICPSR weighted files, to find the candidates, then choose the one that fits the company’s actual activity. Document which NAICS vintage you used.

How do I find my SIC or NAICS code?

Describe your largest source of revenue and search the Census Bureau’s NAICS structure for the matching definition. For a SIC code, check your SEC filing page if you are a public company, or an older government or insurance record. Agencies sometimes assign their own code, so confirm which one a form requires.

Sources

• North American Industry Classification System, U.S. Census Bureau
• 2022 NAICS Manual
• Federal Register notice on NAICS 2022
• SIC and NAICS Codes, Washington Department of Revenue
• CES Published Series, Bureau of Labor Statistics
• Industry Classification Overview, Bureau of Labor Statistics
• Weighted Crosswalks for NAICS and SIC Industry Codes, ICPSR
• Crosswalk files folder, openICPSR
• Statistics Canada, NAICS Canada 2022
• Industry research guide, Library of Congress
• What Is SIC? Standard Industrial Classification

Recommended

• Firmographic Data for B2B: Why It Is Only the First Filter
• Market research
• Data coverage
• Sales & GTM research