A double-blind, randomized, placebo-controlled, parallel group study was performed in 80 adult outpatients meeting ICD-10 criteria for major depression and with a Montgomery-Asberg Depression Rating Scale (MADRS) score of at least 18 at baseline. All patients received paroxetine (20 mg once a day) plus either pindolol (2.5 mg three times a day) or matching placebo for 6 weeks. Analysis of the day 14 MADRS scores on an intent-to-treat basis revealed a treatment-by-centre interaction, with a significant effect of pindolol being demonstrable at only one centre. At this centre, 25% of the paroxetine plus pindolol group and 0% of the paroxetine plus placebo group showed a decrease of at least 50% from baseline MADRS by day 4 (p < 0.05). At day 14, the proportions were 73% and 7%, respectively (p < 0.001). Analysis of covariance on a "perprotocol" population demonstrated a significant accelerator effect of pindolol at days 4 and 7 in the absence of a treatment-by-centre interaction, but a centre effect was apparent at later time-points. The results suggest that the latency of antidepressant action can be reduced with pindolol augmentation. A large multicentre study is in progress to investigate this effect further.