The authors conducted a randomized, double-blind, 10-week clinical trial of two doses of nortriptyline in eight nursing homes. Sixty-nine patients, average age 79.5 years, were randomized to receive regular doses (60 mg-80 mg/day) vs. low doses (10 mg-13 mg/day) of nortriptyline. Among the more cognitively intact patients, there was a significant quadratic relationship defining a "therapeutic window" for nortriptyline plasma levels and clinical improvement. There were also significant differences in plasma level-response relationships between depressed patients who were cognitively impaired and those who were more cognitively intact. Depression remains a syndrome that responds to specific treatment, even in frail nursing home patients, and those depressions that occur in patients with significant dementia may represent a treatment-relevant condition with a different plasma level-response relationship than in depression alone.