We studied the efficacy of nortriptyline (NT) when given in doses which produce tricyclic antidepressant (TCA) levels within the proposed therapeutic "window" in a select patient group to assess the "window" hypothesis in a group of patients that was biologically homogeneous with respect to the TRH test and clinically homogeneous with respect to RDC, DSM III, and Winokur criteria. Pharmacokinetic and dose differences were controlled for by administering a NT dose-prediction test, giving the indicated dose, allowing levels to reach steady state and changing the dose, if necessary, to maintain NT levels within the range of 90-130 ng/ml. Using this protocol nine of ten patients responded to NT. This rate of response for a severely depressed patient group is comparable to response data for ECT and recent European data using regular NT levels to change doses of NT and assess appropriate NT trial duration.