Entrepreneurial activity is often regarded as an engine for economic growth and job creation. Through tax policy, governments possess a potential lever to influence the decisions of economic agents to start and close small businesses. In Germany, the top marginal income tax rates were reduced exclusively for entrepreneurs in 1994 and 1999/2000. These tax reforms provided two naturally defined control groups that enable us to exploit the legislation changes as natural experiments. First, the tax rate reductions did not apply to freelance professionals (Freiberufler), and second, entrepreneurs with earnings below a certain threshold were not affected. Using data from two different sources, the SOEP and the Mikrozensus (LFS), we analyse the effect of the tax cuts on transitions into and out of self-employment and on the rate of self-employment. We apply a difference-in-difference-in-difference estimation technique within a discrete time hazard rate model. The results indicate that the decrease in tax rates did not have a significant effect on the self-employment decision.