Activation measures for unemployed workers such as labor market training typically have negative effects on the employment probability of their participants in the short run. These so-called lock-in effects occur because participants spend less time and effort on job search activities during participation than comparable nonparticipants. Lock-in effects can be quite large, especially for long programs and they need to be offset by sufficiently large positive employment effects after participation for the programs to be cost-effective. They resemble an important indirect cost of activation measures that is often more important than the direct program costs. Therefore, it is important to understand what determines the size of lock-in effects and whether it is possible to reduce them. We study these questions with administrative data for Germany.