S-fluoxetine is the long-acting enantiomer of the racemic antidepressant serotonin reuptake inhibitor. Sixty-five patients needing migraine prophylaxis were recruited into a phase II, double-blind, placebo-controlled trial. After a 1-month placebo run-in, 53 patients met entry criteria with regard to attack frequency and were randomized, 27 to S-fluoxetine and 26 to matching placebo. Three failed to start treatment and there were 17 early discontinuations, 9 from S-fluoxetine, 8 from placebo, at similar times and for similar reasons. The primary efficacy variable was attack frequency and analysis compared decline-from-baseline in the two groups. This was earlier and greater (1.7 attacks/28 days, or 52%) on active therapy than on placebo (1.1 attacks/28 days, or 27%), and statistically significant in month 2 (F = 4.93; p = 0.033) and month 4 (F = 4.55; p = 0.041). As secondary measures of efficacy, migraine-days per month and Patient's Global Impression of Disease Severity coherently reflected the changes in attack frequency. Mean attack severity and acute medication use (doses per attack) were unaltered by either treatment. There were no serious adverse events. Withdrawals for adverse events were four from each group but none was considered causally related. The finding of greater efficacy of S-fluoxetine than of placebo should be interpreted conservatively, since the analysis in the final month was made on only half of the entered patients. It supports progression to phase III evaluation, which was the purpose of the study.