Using data from a large-scale randomized controlled trial conducted in Danish job centers, this paper investigates the effects of activating sick-listed workers on subsequent labour market outcomes. Comparing treated and controls, we find an overall unfavourable effect on subsequent labour market outcomes. Using variations in activation regimes (both between job centers and between randomly assigned treatment and control groups within a given job center) as an instrument for participating in a specific activity, we compare the relative effectiveness of alternative activation strategies. Our results show that the use of partial sick leave increases the length of time spent in regular employment, and also reduces the time spent in unemployment and in early retirement. Traditional active labour market programs and the use of paramedical care appear to have no effect at all, or even an adverse effect.