It is now widely accepted that tricyclic compounds are of value in a proportion of depressed patients and that drugs of the imipramine group can take four or more weeks to exert their effect (Moody et al., 1967). There is a temptation to assume that one must wait this period before assessing response and that this is necessarily the best time for predicting the ultimate outcome of treatment. In experience with amitriptyline and nortriptyline both authors gained the impression that patients who were going to respond well were already a clinically distinct group after two weeks' exhibition of the drug, and indeed that waiting till four weeks blurred the distinction because by then poor responders might also have improved. It was decided to test this impression further.