There is a lack of evidence about what is the best treatment for people who present to hospital after self harm. Most treatment trials have been small and involved unrepresentative groups of patients which result in inconclusive findings. Here we note some of the characteristics of attempted suicide which make it a difficult subject to study. We describe the problems of doing randomized controlled trials in attempted suicide and outline the advantages and difficulties of randomized controlled trials, Zelen designs, patient preference designs, and cluster randomized trials in attempted suicide intervention trials. Researchers and consumers should consider other research designs when asking what is effective after self harm.