Patients with a diagnosis of alcohol dependence, detoxified and abstinent for 10-30 days, were randomly allocated to placebo or the serotonin reuptake inhibitor, fluvoxamine (up to 300 mg per day), plus counselling and support. In the intention to treat sample of 493, there was a trend for the fluvoxamine group to do worse than the placebo group on the primary outcome criteria: abstinence; and relapse defined as drinking > or =5 units on an occasion and > or =4 such occasions in a week, or > or =12 units on an occasion (1 unit = 9g ethanol). When typology of alcoholism was assigned by scores on the Tridimensional Personality Questionnaire, Types I and II had similar rates of survival without relapse on placebo (PLC I: 19.3%, n = 135; PLC II: 18.2%, n = 110), but on fluvoxamine Type II did worse than Type I (FLU I: 13.7%, n = 131; FLU II: 6.14%, n = 114) (P < 0.01). When typology was assigned on the basis of age of onset of alcohol problems (< or = age 25, or > age 25), early-onset patients in the fluvoxamine group relapsed more frequently than late-onset patients in that group (no longer significant after adjustment for gender), as did those who commenced regular drinking before age 25 (both with and without adjustment for gender). One explanation for our finding could be that impulsivity in early-onset or Type II patients may be accentuated by serotonin enhancement.