OBJECTIVE: To investigate the degree of ejaculation delay induced by on-demand treatment with 20 mg paroxetine and 25 mg clomipramine and to assess the type and severity of non-sexual side-effects of treatment at the day of and the day after treatment with these drugs. METHOD: A randomized, double-blind, fixed-dose, on-demand study in 30 men with lifelong premature ejaculation was performed. During a 1-month baseline period and a 4-week drug treatment period patients assessed the intravaginal ejaculation latency time (IELT) at home with a stopwatch. Only men with an IELT <1 min were randomly assigned to drug treatment. Patients assessed the drug coitus interval time (DCIT) and used the UKU side effect scale questionnaire at baseline, the day of and the day after intercourse. RESULTS: On-demand treatment with 25 mg clomipramine, with a mean DCIT of 5.14 h, led to a 4.05 (95%CI: 3.26-5.02) fold-increase of the IELT. On-demand treatment with 20 mg paroxetine, with a mean DCIT of 5.39 h, led to a 1.41 (95%CI: 1.22-1.63) fold-increase of the IELT. Both drugs had a high incidence of non-sexual side effects at the coitus day and the next day. At the day of coitus paroxetine led to significant sleepiness and yawning compared to clomipramine. At the day after coitus clomipramine induced significant nausea compared to paroxetine. CONCLUSION: On-demand treatment with 25 mg clomipramine led to a clinical relevant ejaculation delay. In contrast, 20 mg paroxetine had no clinical relevant ejaculation delay in men with lifelong premature ejaculation with an IELT of less than 1 minute. Both drugs exert mostly mild yet annoying non-sexual side effects both at the coitus day and the next day.