The causal impact if starting atypical work on the return to regular employment is measured using the timing-of-events approach. The employment and unemployment history of 1/25th of French workers was reconstructed using three linked administrative data sources. During unemployment, starting atypical work is found to raise the likelihood of finding regular work by 75% in the following months, a robust stepping-stone effect. We find no evidence atypical work on wage growth and no lock-in effects. Long term unemployed workers, older job seekers, and those who did not work in the months before starting to look for a job have a lower chance of entering atypical work, but they benefit more from doing so.