In a controlled trial, the beta-adrenoceptor/5-hydroxytryptamine-1A (5-HT1A) receptor antagonist pindolol accelerated and enhanced the antidepressant effect of fluoxetine. The median times to sustained response (> or = 50% reduction of baseline severity maintained until endpoint) were 19 days for fluoxetine plus pindolol (N = 55) and 29 days for fluoxetine plus placebo (N = 56) (p = 0.01). The response rate at endpoint was 16% greater in patients treated with the combination. The plasma concentration of pindolol remained stable between 3 days (first blood sampling) and 6 weeks. Mean values were approximately 26 nM, a concentration higher than the Ki of (-)pindolol for human 5-HT1A autoreceptors (11 nM). Plasma fluoxetine and norfluoxetine concentrations increased steadily until the fourth week of treatment. Fluoxetine concentrations were lower in patients receiving the combination (p = 0.043), but there was no significant relationship to the clinical response in either group. A reanalysis of the data using a survival analysis revealed that significant differences in the time to sustained response between both groups would have also been detected (1) in a 2-week trial, (2) without a placebo lead-in phase, and (3) with less frequent visits. However, the use of "response" instead of "sustained response" as measure of clinically relevant change would have greatly diminished the difference between treatment arms (p = 0.08 instead of p = 0.01). This emphasizes the need of using stringent outcome criteria in antidepressant drug trials. A comparison of the data of all sustained responders (N = 27) in the fluoxetine-plus-placebo group with the first 27 responders in the fluoxetine-plus-pindolol group (of a total of 38) revealed a highly significant difference in the time to sustained response (18 and 10 days, respectively; p = 0.0002). This indicates that the faster response in the fluoxetine-plus-pindolol group is not a result of the greater proportion of responders.