Many papers have already been published on the technique of insulin therapy for the psychoses (Müller, 1936; Sakel, 1937 and 1938; Frostig, 1937; Küppers, 1937; Braunmühl, 1938, and others). Nevertheless, some of the details of the method still offer problems to the careful therapist who wishes to avoid impairing his therapeutic effects by excessive caution, but is unwilling to incur unpredictable risks. While investigating this therapy during 1938 at the Maudsley Hospital, we endeavoured to evolve a technique which was both effective and safe, and which could be individualized to each patient without thereby losing its precision. In the following paper we offer suggestions concerning some of the technical details that have appeared to us important in achieving these ends. As far as possible we have omitted points on which previous papers have given adequately clear and consistent instructions.