Most welfare programs generate high marginal tax rates on labor income. This paper uses a representative sample of individuals on France's main welfare program (the Revenu Minimum d'Insertion, or RMI) to estimate monetary gains to employment for welfare recipients. This is based on the distribution of potential monthly earnings faced by each individual, as inferred from the distribution of observed wages and working time. Taking account of the welfare earnings top-up program (intéressement), we find that gains are almost always positive, but that their amount is very low, especially for single mothers. Intéressement is found to have a small impact, because of its provisional nature. Gains are positively related to the probability that a welfare recipient in 1996 will be observed in employment in 1998. Using a simple structural model, we interpret this as a labor supply effect.