The authors investigated the relationship of desipramine concentrations in plasma to response, side effects, and dose in depressed patients over 75 years of age to determine if these "very old" patients were unusually sensitive to treatment. Thirty-four elderly patients consecutively hospitalized for nonpsychotic, unipolar major depression were treated with a fixed dose desipramine regimen for 4 weeks. Twelve nonresponding patients received a second trial at an increased dose. Comparisons were made with data from younger patients previously published by the authors. At comparable doses, steady-state desipramine concentrations in plasma in the elderly patients did not differ from those observed in younger patients. Response at levels in blood < 115 ng/ml was low, only 6 (21%) of 28 patient trials resulted in response. At levels > or = 115 ng/ml, 6 (46%) of 13 patient trials were effective. These rates were not significantly different. Inspection of the data revealed that a concentration in plasma of 105 ng/ml significantly separated responders and nonresponders (chi 2 = 3.93, df = 1, p < 0.05), but even at levels > or = 105 ng/ml, the response rate was still low relative to rates in prior studies of younger patients treated for a similar duration. The serious adverse reaction rate, 7 of 34, was similar to that previously observed in younger patients. This sample of "very old" elderly was not unusually "sensitive" to antidepressant drug treatment. In fact, the low rate of response observed at usually adequate levels in blood suggested "resistance" to treatment. The findings underscore the need for more effective drug treatments in the depressed elderly.