Availability of (partial) insurance mechanisms is arguably important for the decision of (riskaverse) workers to start up a risky entrepreneurial venture. Using administrative data from Denmark, where unemployment insurance (UI) is available to both wage earners and self-employed on a voluntary basis, we estimate the causal effect of UI cover on the self-employment choice of wage earners after instrumenting for the UI choice. The instruments we use are based on a series of policy variations that took place at three points in time during an observation period spanning three decades: only UI covered individuals could under certain conditions qualify for an early retirement (ER) program. Changes (reforms) in the eligibility conditions of the program that affected different age groups differentially at these three different points in time identify the UI choice process. Results show that the causal effect of insurance on the probability of starting up a venture is positive for would-be entrepreneurs, in contrast to correlations in the data or uninstrumented estimates. Using firm data, we also investigate how the newly insurance-induced entrepreneurs fare relative to their uninsured peers. Results suggest that they survive longer, but are not more likely to employ any workers or to make higher or lower profits.