Federal data on United States businesses shows that about four in five new establishments survive their first year and about half survive five, though the figures describe locations rather than company founders’ fortunes. This article explains what the Bureau of Labor Statistics (BLS) numbers measure, how stable they have been, and how to avoid over-reading them. It describes United States data only; other countries measure business survival differently and report different results.
What the BLS survival tables actually measure
The BLS publishes survival statistics through its Business Employment Dynamics program, which it describes as a set of statistics generated from the Quarterly Census of Employment and Wages. The program’s quarterly series cover gross job gains and gross job losses from 1992 forward. One of its tables, titled “Survival of private sector establishments by opening year,” follows each yearly group of newly opened private-sector establishments and reports the share still operating one, two, five and ten years later.
The unit is the establishment, which means a business location with employees on the payroll, rather than a company or a founder. That matters for interpretation. A second branch opened by an existing firm is a new establishment. A business that closes one location and moves to another may be counted as a closure and an opening. The data also cover employer businesses in the private sector, so a person working alone with no employees is outside the picture.
The cohorts are defined by year ended March. A 2015 cohort, for example, is read at March of each later year, so its ten-year rate is measured in 2025.
How many new establishments survive one, two, five and ten years?
According to the BLS table, the cohorts that opened in the years ended March 1994 through 2015 show the following survival rates since birth, in percent:
| Opening cohort (year ended March) | 1 year | 2 years | 5 years | 10 years |
|---|---|---|---|---|
| 1994 | 79.6 | 68.1 | 49.6 | 33.6 |
| 2005 | 80.1 | 68.7 | 46.8 | 33.8 |
| 2010 | 78.6 | 68.6 | 51.1 | 35.3 |
| 2015 | 79.6 | 69.1 | 50.2 | 34.7 |
Across every cohort in the table, the one-year rate ranges from about 75% to about 81%, the two-year rate from about 63% to 72%, and the five-year rate from about 45% to 52%. For cohorts with ten years of history, the ten-year rate sits between about 32% and 35%.
Put plainly, as an average pattern in US data, a little under half of new establishments are gone within five years, and roughly two-thirds are gone within ten. These are population averages across all private industries, not a forecast for any single business.
Is the failure rate getting better or worse?
The striking feature of the table is how little the headline rates have moved. The 1994 cohort had a five-year rate of 49.6%. The 2015 cohort had 50.2%. The ten-year rate was 33.6% for 1994 and 34.7% for 2015, a difference of 1.1 percentage points over two decades.
The movement that does exist tracks the economy. Cohorts opened around the 2008 financial crisis show lower early survival: the cohort for the year ended March 2008 had a one-year rate of 75.2% and a two-year rate of 63.3%, both among the lowest in the table. Cohorts from 2020 show the opposite, with a one-year rate of 80.9% and a two-year rate of 72.3%, the highest values in those columns. The BLS table does not explain why. One reasonable reading, offered as analysis rather than something the data prove, is that conditions at the time of opening, including demand and the availability of credit, influence early survival, while the longer-run rates settle toward a similar level.
More recent cohorts are incomplete. The 2022 cohort shows 76.3% after one year and 64.8% after two, and five-year figures for cohorts opened after 2020 do not yet exist.
When do establishments close: early or late?
The cumulative rates hide a pattern that a simple calculation reveals. Dividing each later rate by the earlier one shows how many of the survivors at one point are still around at the next. Using the 2015 cohort as a worked example:
- 79.6% survived one year.
- 50.2% survived five years. Dividing 50.2 by 79.6 gives about 63%, so roughly 63% of first-year survivors lasted to year five.
- 34.7% survived ten years. Dividing 34.7 by 50.2 gives about 69%, so roughly 69% of five-year survivors lasted to year ten.
An illustrative cohort of 1,000 new establishments with those same rates would look like this:
| Point in time | Establishments still open | Closed since previous point |
|---|---|---|
| Opening | 1,000 | n/a |
| After 1 year | about 796 | about 204 |
| After 2 years | about 691 | about 105 |
| After 5 years | about 502 | about 189 |
| After 10 years | about 347 | about 155 |
The table is illustrative arithmetic based on the published rates, not additional data. It shows that the first year is the most hazardous single period, with about one in five closing, and that the risk eases with age. Closure does not disappear, though. Even among the group that reaches year five, about three in ten are gone by year ten on these numbers.
What does a closure mean?
Closing an establishment is not the same as failing. The BLS survival data record that an establishment stopped appearing in payroll records. They do not record why. Possible reasons include insolvency, but also retirement, a sale, a merger, a lease ending, a decision to relocate or an owner choosing to do something else. The data cannot separate a bankruptcy from a planned wind-down.
This is also why survival statistics should not be compared directly with insolvency statistics. For the England and Wales system, for example, reading UK company insolvency figures shows that a formal insolvency count captures only companies that went through a legal procedure, and most were orderly closures. The two series answer different questions in different countries, and neither is a measure of how many owners lost money.
How big is the churn behind the survival rates?
Survival rates make more sense next to the scale of openings and closings in the wider economy. The BLS release for the fourth quarter of 2025, published on July 29, 2026, reported the following seasonally adjusted private-sector job flows:
| Measure | Jobs |
|---|---|
| Gross job gains | 7,838,000 |
| of which expanding establishments | 6,197,000 |
| of which opening establishments | 1,641,000 |
| Gross job losses | 7,245,000 |
| of which contracting establishments | 5,812,000 |
| of which closing establishments | 1,433,000 |
| Net employment change | +593,000 |
The components add up: 6,197,000 plus 1,641,000 equals 7,838,000, and 5,812,000 plus 1,433,000 equals 7,245,000. The difference between gains and losses is the net change of 593,000. According to the release, gross gains equaled 5.9% of private-sector employment and gross losses 5.5%.
Two points follow. First, in a single quarter, openings added more jobs than closings removed (1.6 million against 1.4 million). Second, contractions and expansions at existing establishments were several times larger than openings and closings. Most of the movement in employment comes from existing businesses growing and shrinking, not from births and deaths alone. The release also reported 338,000 establishment births in the quarter, accounting for 1,006,000 jobs, with death data available only through the first quarter of 2025, when 284,000 establishments closed and 809,000 jobs were lost.
Why the averages can mislead an individual business
An average across all industries and locations blends very different situations. A restaurant, a software consultancy, a construction contractor and a branch of a national retailer carry different risks, capital needs and customer bases, and the table shown here covers total private-sector establishments only. Industry-specific results can differ widely from the all-industry figure, so the 50% number is a starting reference, not a benchmark for any particular sector.
The data also say nothing about size or financing. Whether an establishment opened with strong cash reserves or on a thin margin is not captured. Cash timing is one factor a business owner can manage, as explored in why cash flow discipline matters more than revenue growth, but the survival table itself does not test any explanation for closures.
Lastly, establishments that survive may be unprofitable, while some closures follow a deliberate sale at a gain. Survival is a count of existence, not a measure of success or owner returns.
Where will the next update come from?
The BLS has scheduled the first quarter 2026 Business Employment Dynamics release for Wednesday, October 28, 2026, at 10:00 a.m. Eastern Time, and says it will incorporate annual revisions. Anyone tracking these figures should expect earlier numbers to be revised, since the data derive from administrative payroll records that are updated over time. For readers following how smaller employers cope with a tougher environment, how small and mid-sized businesses are adapting to higher costs offers context that the survival tables, by design, cannot.
Reading survival data with the right expectations
The federal record shows a consistent pattern: about 80% of new private-sector establishments reach their first anniversary, about half reach their fifth and about a third reach their tenth, with little change across cohorts since the 1990s. The numbers describe locations, not founders; closures, not necessarily failures; and averages, not odds for any single business. Used with those limits in mind, they offer a grounded reference point against which anecdotes and headline claims about startup failure can be checked.
This article is general information, not personalised financial advice; rules, statistics and rates change.
Sources: US Bureau of Labor Statistics
