This paper evaluates the effectiveness of the training programs offered to the unemployed in Wallonia, the French-speaking part of Belgium. More precisely, we are interested in the two following questions : (a) does training increase the job finding rate of the unemployed after the completion of their training ? (b) if so, is this increase big enough to compensate the so-called lock-in effect of their training ? To answer these questions, we rely on the Abbring and van den Berg (2003) timing-of-events approach and a very large administrative dataset. We find that training has an overall strong and persistent effect on the job finding rate of the unemployed after the completion of their training, but that this effect is heterogeneous, varying according to the characteristics of training and trainees. Further, we find that this effect on the job finding rate is globally large enough to compensate the lock-in effect of the training, but again with significant heterogeneity.