How Long Do Businesses Last? BLS Survival Data Explained

What survival data measure

"How many businesses survive five years?" has a standard public answer, and it comes with a definition most people skip. The U.S. figure that SBA, journalists and pitch decks quote comes from the Bureau of Labor Statistics Business Employment Dynamics (BED) program, and it measures establishments, not firms and not owners.

Three definitions from BLS govern every number in this article:

• Age. "The age of an establishment is determined by its date of first positive employment" ^[1].
• Opening. Openings are establishments with positive employment in the current year that had no employment in the prior March ^[1]. Cohorts run on years ending in March.
• Coverage. Age data "are available from March 1994 through March of the current year" ^[1]. The survival tables are Tables 5 to 7: Table 5 counts surviving establishments from each opening cohort, Table 6 their employment, and Table 7 "profiles the establishment survival rates for each cohort by opening year" ^[1].

An establishment is a single location with employees. A multi-location company opening a new branch creates a new establishment; BLS notes that the same event is an opening under the establishment method and an expansion under a firm method ^[1]. The data cover employer establishments only. Businesses with no employees are not in BED.

The headline numbers

The SBA Office of Advocacy summarizes the BED series in its annual FAQ. The 2026 edition states: "From 1994-2022, an average of 67.7 percent of new employer establishments survived at least two years. During the same period, the five-year survival rate was 49.2 percent, the ten-year survival rate was 33.9 percent, and the fifteen-year survival rate was 25.5 percent" ^[2].

The same answer adds the conditional figures that matter for acquirers: "more than two thirds of businesses (69.5 percent) that reach five years reach ten", and "76.1 percent of establishments that reach ten years also reach fifteen" ^[2].

Source: SBA Office of Advocacy, FAQ 2026, citing BED. Earlier editions of the FAQ cover different year ranges and give slightly different averages; cite the edition and its period.

In plain terms: about half of new employer establishments close within five years, and the risk of closure falls steeply for those that survive.

Cohort table: how stable are the rates?

Averages hide variation between cohorts. BLS Table 7 gives the survival path of every annual cohort since 1994. Selected cohorts, read from the raw table on 2026-10-28:

Source: BLS BED Table 7, total private. "n/a" means the cohort has not yet reached that age in the data.

Across the 28 cohorts that have reached age five (1994 through 2021), the five-year rate ranges from 45.4% (2006 cohort) to 51.9% (2018 cohort). Across the 23 cohorts that have reached age ten, the ten-year rate ranges from 32.4% (2001 cohort) to 35.3% (2010 cohort) (BLS BED Table 7; ranges computed from the table). The three lowest five-year rates belong to the 2005, 2006 and 2007 cohorts (46.8%, 45.4% and 46.4%), whose first five years ran through the 2008-2009 recession. The series is stable enough to use as a baseline, but it is not flat.

The oldest cohort is now 32 years old. Of the 569,387 establishments that opened in the year ended March 1994, 68,378 were still operating in March 2026, a survival rate of 12.0% ^[3].

Conditional survival by age

Table 7 also reports the "survival rate of previous year's survivors", the share of establishments alive at the start of a year that are still alive at its end. Averaging that column across all cohorts that have reached each age gives a simple picture of how risk changes with age:

Computed by this article from BLS BED Table 7 as unweighted means across cohorts; BLS does not publish these averages.

The steepest drop is in the first two years. After about ten years, roughly 94 to 96 percent of surviving establishments make it through each additional year, and the rate keeps improving only slowly.

Establishment vs firm vs owner

Three different things can "survive", and the public data measure only the first two:

The Census Bureau's Business Dynamics Statistics (BDS) covers the firm dimension: it provides "establishment births and deaths, and firm startups and shutdowns", with tables on data.census.gov "for all years, 1978 to 2023" ^[4]. BDS and BED use different sources and definitions, so do not mix their rates in one table. The same caution applies to the business counts compared in CBP vs QCEW vs SUSB.

Neither series tells you whether the founder is still the owner. A 25-year-old establishment may have changed hands twice. For an origination team that distinction matters more than any survival rate; see succession and ownership for the evidence types that bear on it.

What survival does not tell you

• Closure is not failure. BED records when an establishment stops having employees. That can follow a sale, a merger into another location, a relocation, a retirement or a voluntary wind-down, as well as insolvency. The BLS FAQ notes that establishments involved in breakouts or consolidations within multi-unit firms carry the predecessor's date of first positive employment ^[1]; it does not describe ownership changes as closures, and survival rates should not be read as rates of business failure.
• Industries differ. BLS publishes the same age and survival tables for each major NAICS sector as well as for total private employment ^[5]. Use the sector table for the industry you screen, not the national total.
• States differ in coverage. State data are available only at the total private level ^[1].
• Survival is not quality. A long-lived establishment has passed the early-closure years. That says nothing about margins, customer concentration, or whether the business depends on one person.

How to state survival figures correctly

Survival statistics are among the most misquoted numbers in small-business writing. The common errors, and corrected wording:

Two habits prevent most of these errors: always name the unit (establishment) and the period, and never apply a population rate to a single target as if it were a probability for that company.

Implications for acquisition screening

Most lower-middle-market screens include an age criterion, often "in business 10+ years". The data support a narrow reading of what that criterion does:

1. It removes most early-stage risk. Conditional one-year survival rises from about 79% in the first year to about 94% by year ten in the cohort averages above. An establishment past ten years has survived the steepest part of the curve.
2. It does not select for quality beyond that. Past ten years, year-to-year survival improves slowly. A 30-year-old business is not much less likely to close next year than a 12-year-old one, by this measure.
3. Age is a weak proxy for owner age. Screens sometimes use founding year as a signal of succession timing. A company founded in 1995 could be run by its founder, a second-generation owner, or a buyer who acquired it last year. Treat founding year as a prompt to look for ownership evidence, not as evidence itself.
4. Use the right founding date. BED age is date of first positive employment at that location. A company's own "established" claim may be the founding of a predecessor, a brand or the owner's first trade license. Record which date you are using and its source.

Write the criterion with its definition. "Operating at least 10 years, measured from the earliest of: state registry formation date, first dated public record, or company's own founding claim (recorded separately)" is auditable. "Established business" is not. For how a criterion like this sits inside a full screen, see screening criteria.

How to pull the tables yourself

1. Open the BLS establishment age and survival data page and choose total private or the NAICS sector that matches your thesis.
2. Open Table 7 for that sector. Each block is one opening cohort; rows are the March of each subsequent year.
3. Read "Survival rates since birth" for cumulative survival and "Survival rates of previous year's survivors" for conditional one-year survival.
4. Record the cohort, the horizon, the sector and the date you retrieved the table. BLS updates the tables annually and the most recent cohorts gain one more year each time.
5. If you compare industries, use the same cohorts and horizons for each.

Worksheet

The construction values in the right-hand column are invented to show the format; pull the real NAICS 23 table before using any number.

Caveats

• Release timing. Table 7 now runs through March 2026. The most recent cohorts have only one to five years of history, so their longer-horizon rates appear as missing.
• Cohort years end in March. A "2010 cohort" opened between April 2009 and March 2010.
• Averages and editions. The SBA averages cover 1994-2022 in the 2026 edition; computed ranges here use all cohorts available in Table 7 on the retrieval date. State which you use.
• Establishment level throughout. None of these figures is a firm survival rate or an owner tenure rate.

Docket's Triage agent resolves a company's identity and legal entity before research is paid for, which is where founding-date evidence for an age criterion is first recorded.