We analyze the effectiveness of publicly financed training and retraining programs in east Germany as measured by their effects on individual re-employment probabilities after training. These are estimated by discrete hazard rate models on the basis of individual-level panel data. We account for unobserved individual heterogeneity in both the training participation and outcome equation. The latter differentiates between transitions into "stable" and "unstable" employment after the completion of a training program. Our findings are that in the first phase of the east German transition process, when the institutions delivering the training programs were being set up, there are no positive effects of training on the probability to find stable employment. For the period of September 1992 to November 1994, when the institutional structure for the programs was in place, we find positive effects of both on-the-job and off-the-job training for women, and positive effects of off-the-job training for men.