Aims: To study the pharmacokinetics of fluvoxamine when given in increasing doses to healthy volunteers., Methods: Ten healthy, non-smoking men were given maintenance treatment with fluvoxamine for 4 weeks. Eight subjects were CYP2D6 extensive metabolisers (EMs) and two were CYP2D6 poor metabolisers (PMs). As a measure of the CYP1A2 phenotype, the paraxanthine/caffeine ratio in saliva after intake of caffeine was studied. The fluvoxamine doses given were 25 mg day(-1) the first week, 50 mg day(-1) the second week, 100 mg day(-1) the third week and 200 mg day(-1) the fourth week, divided in two daily doses. On the seventh day every week, serum concentrations of fluvoxamine were followed for a dose interval of 12 h. After discontinuation of treatment, fluvoxamine concentrations were followed for 1 week., Results: For each of the three two-fold increases in given dose, the mean AUC increased 3.25-fold, 3.17-fold and 3.14-fold, respectively (P < 0.0001), indicating a decrease in oral clearance with increasing dose. The elimination half-life based upon the serum concentrations 12-48 h after discontinuation of fluvoxamine was 32.1 +/- 11.0 h whereas the half-life based upon the concentrations 3-7 days after discontinuation was significantly shorter, 15.8 +/- 4.2h (means +/- s.d.; P < 0.001). There were no significant correlations between the CYP1A2 phenotype and fluvoxamine AUCs at different doses (r = -0.56; P = 0.095 for the correlation between the paraxanthine/caffeine ratio in saliva and fluvoxamine AUC at a dose of 50 mg day[-1]). The two CYP2D6 PMs had AUC values in the same range as the EMs., Conclusions: The present study conclusively demonstrates that fluvoxamine exhibits non-linear kinetics within the therapeutic dose interval. The reason for non-linearity is not Michaelis-Menten saturation kinetics of a single metabolic pathway, but rather a complex involvement of multiple parallel pathways.