[Fitting proof]

Randomised controled trials (RCT's) and meta analyses of RCT's are known as the best research designs to evaluate if interventions are doing more good than bad. Some interventions can not be evaluated by RCT's because of the heterogeneity of the problems, the cost of the evaluation study or ethical arguments against the study. This is often the case with population based interventions. A typical example is suicide prevention. The shortcomings of the "classic" research designs for the evaluation of suicide prevention are discussed and feasible solutions are suggested for future research.