Psychiatry and psychopharmacology are no longer aiming to make a decisive breakthrough at the end of the century. Rather than seeking explanations, research workers are looking for ‘predictions’. Three main types of prediction are emerging: a tautological, a heuristic, and an irrelevant one. Few predictions found in the recent literature can be marked as ‘logical’ ones. Nevertheless, predictions play two important roles: they generate new hypotheses that can be falsified in properly designed scientific experiments; they also may serve to falsify given hypotheses. The main recent findings on predictions in psychiatry are briefly summarised.