Back in 2008, I wrote a post on startup failure rates  that has proved to be very popular. Because that post was based on the 1992 cohort of new companies, many people have asked me if the numbers are true for today’s start-ups. Roughly speaking, the answer is yes.
But don’t take my word for it. Look at what the U.S. government data show. In the chart below, I present the latest available numbers from the two government statistical agencies responsible for providing data about new businesses: The Census Bureau and the Bureau of Labor Statistics  (BLS).
The chart includes both new establishments, which the BLS defines as new economic entities doing business at a particular location, and new firms, which it defines as new companies. While new firms may have more than one establishment and new establishments may be started by existing companies, researchers have found that the number of new firms and the number of new establishments are similar.
While you could find these numbers directly from the original sources, you won’t find them together. Moreover, they don’t appear very high in Internet searches. Because the older and less accurate numbers are easier to find than the newer and better ones, lots of inaccurate information about new business failure rates is constantly being reproduced and reused.
The figure below shows six sets of numbers. The longest series, the 1994 BLS EST line, shows the percentage of new establishments founded in 1994 still alive in every year through 2010. The figure below the curve is the percentage of companies started in 1994 still alive that many years after founding. While the 1994 cohort is old, you can’t get the survival rate of businesses in their 17th year by looking at more recent cohorts.
The 1995 BLS EST, 2000 BLS EST, and 2005 BLS EST lines each track the five year survival rates of the cohorts of new establishments founded in 1995, 2000 and 2005 respectively. Five years after they were started, 50, 49 and 47 percent of the new establishments started in 1995, 2000 and 2005, respectively, were still alive.
The remaining two series come from the Census Bureau. The 2005 CENSUS EST line shows the survival rate of new establishments founded in 2005 through 2010, while the 2005 CENSUS FIRMS line shows the survival rate of new firms started in 2005 over the same period. Five years after they were started, the Census Bureau finds that 45 percent of the new establishments and 43 percent of the new firms were still alive.
While I have thrown a whole lot of numbers at you, I am making a very simple point: The typical new business started in the United States is no longer in operation five years after being founded. That’s true whether statisticians at the BLS or Census are doing the measuring and it’s true whether you measure new establishments or new firms.
Startup Survival Rates
Source: Created from data from Longitudinal Business Database 1977-2010, Census; Business Employment Dynamics 1994-2010, Bureau of Labor Statistics