A meta-analysis was carried out to evaluate data published between January 1974 and February 1998 comparing rates of treatment response and tolerability of pharmacological and psychological treatments for depression in persons over age 55. Drugs (tricyclic antidepressants, selective serotonin-reuptake inhibitors, and a mixed group of other drugs) were significantly better than placebo, with an average reduction in symptom severity of 48.0% versus 31.3% (analysis weighted by sample size; 50.6% vs. 21.4% unweighted). No single drug or group of drugs was superior in terms of efficacy, and no statistically significant differences in tolerability emerged between tricyclic antidepressants and selective serotonin-reuptake inhibitors, whether measured by total dropouts or by dropouts due to side effects. Compared to the data on pharmacological treatments, those for outcomes of psychological treatments are very limited. Existing data indicate that cognitive-behavioral, behavioral, and psychodynamic therapies are significantly better than placebo. In the current meta-analysis, response rates to these nondrug therapies did not differ significantly from those observed with tricyclic antidepressants or selective serotonin-reuptake inhibitors, but direct comparison data are insufficient for firm conclusions to be drawn about comparative efficacy. It is possible, even likely, that not only different subtypes of depression but also different patients vary in their treatment responses. However, lack of adequate data prevented the current meta-analysis from addressing these issues. Rigorously designed prospective studies on treatment outcome, taking into account the above differences, are urgently needed to provide robust data on which to base clinical recommendations for the treatment of depression in older patients.