Severe pruritus may be an idiopathic phenomenon or associated with advanced systemic disease. It is one of the most distressing and difficult to treat symptoms. Uncontrolled studies have suggested that, in patients experiencing severe pruritus, paroxetine appeared to have a rapid anti-pruritic effect. This study was a prospective double-blind, randomized within patient comparison of paroxetine and placebo. The intensity of pruritus was measured subjectively with a numerical analogue scale. The primary endpoint of the trial was the mean pruritus score, measured for seven days after randomization and after cross-over. The secondary endpoint was individual global response to the treatment. Response was defined as at least 50% reduction of intensity of pruritus in the last three days of the treatment period vs. baseline. Adverse effects and patient satisfaction and preferences were also recorded. Twenty-six patients were included in the study; 17 of them had solid tumors, 4 had hematological malignancies and 5 had various nonmalignant or idiopathic conditions. Eight patients had drug-induced pruritus (none opioid-induced), 7 patients had paraneoplastic pruritus and 3 had cholestatic pruritus. After a run-in period, patients were randomly assigned to treatment with 20 mg paroxetine or placebo. The crossover took place after 7 days. Two patients discontinued treatment because of adverse effects of paroxetine. Twenty-four patients treated with paroxetine had lower pruritus intensity scores over the 7 treatment periods (mean+/-SE=5.2+/-0.32) as compared to placebo (mean+/-SE=6.0+/-0.32). Mean difference between placebo and paroxetine was 0.78 (95% CI=0.37-1.19). Nine of twenty-four patients (37.5%) fulfilled criteria of response. The onset of anti-pruritic action was observed usually after 2-3 days, irrespective of the order of treatment. The outcome of this study indicates that paroxetine is effective in the treatment of severe pruritus of non-dermatological origin.