Adolescents with major depressive disorder (N = 25) were treated with nortriptyline 100 mg for 6 to 10 weeks. Clinical change was assessed by three treatment outcome measures: the Schedule for Affective Disorders and Schizophrenia (K-SADS-III-R) (to assess syndromal recovery, i.e., no longer fulfilling diagnostic criteria for the disorder), the self-rated Beck Depression Inventory, and a clinician-rated 17-item Depression Scale (derived from the K-SADS-III-R). At 6 weeks, recovery rates for full response were 44% to 57%, for partial response were 26% to 32%, and for nonresponse were 12% to 28%. Better outcome was not associated with concurrent treatment (hospital or outpatient), endogenous or nonendogenous status, or pretreatment depression severity. A significantly greater number of full and partial responders had plasma nortriptyline levels in the adult range of 50-150 ng/ml. None of the response measures, when analyzed as continuous variables, were curvilinearly associated with plasma nortriptyline levels, a finding which does not support the therapeutic window hypothesis. The 17 adolescents who were treated for 10 weeks continued to improve on all three outcome measures (syndrome, p < 0.07; Beck, p < 0.05; 17-Item, p < 0.004) relative to their response at 6 weeks. By 10 weeks, all three measures suggested clinical improvement, but no conclusions about efficacy can be offered in view of the open-label design, absence of placebo controls, diagnostic subtype heterogeneity, small sample size, and concurrent treatments.These data suggest that apparent improvement in adolescent major depressive disorder may vary, depending on treatment outcome criteria, plasma nortriptyline level, and treatment duration. Clinical recovery appears more rapid when treatment response is defined by the clinician-rating (17-Item Depression Scale) and slower when defined by self-rating (Beck Depression Inventory) or by diagnostic status (syndromal criteria). The clinician-rated scale may be more sensitive to early improvements in adolescents, as in adults. Extending treatment from 6 to 10 weeks increased the response rate by about 50%, so a 10-week trial of nortriptyline might be clinically indicated before nonresponse status in adolescents is established.