In this paper, we evaluate the employment effects of Danish active labour market programmes aimed at welfare benefit recipients. We estimate an econometric model with treatment effects and discrete outcomes and we allow the responses to treatment to vary among observationally identical persons. The empirical analysis is based on a register-based dataset that gives information on participation in labour market programmes and subsequent employment. Using a latent variable model, we estimate commonly defined treatment effects, and in particular, the distribution of treatment effects. We do not find any significant mean effects of participation in private sector employment programmes compared to participation in other programmes, but we find evidence of heterogeneity in the treatment effects.