BACKGROUND Assessment of the relation between oral risperidone dose, serum drug levels and clinical response may provide important information for rational treatment decisions. Inter-individual differences in the liver cytochrome P450 system, especially in the CYP2D6 subsystem, which account for a significant portion of risperidone metabolism, may also influence plasma drug levels and alter clinical response parameters. We thus prospectively investigated risperidone serum concentrations in relation to clinical efficacy and side-effects and genotyped major CYP2D6 polymorphisms to determine their effect upon these parameters. METHODS Neuroleptic monotherapy with risperidone was administered to schizophrenia patients in a 6-week open dose clinical trial. Weekly assessments including CGI and PANSS ratings to assess psychopathology; SAS to assess medication side effects; and blood draws to quantify steady state plasma levels of risperidone and 9-OH-risperidone were carried out. In addition, major CYP2D6 polymorphisms including alleles *4, *6 and *14 were genotyped. RESULTS Eighty-two patients were recruited. Mean oral dose of risperidone was 4.3 +/- 0.9 mg. Mean plasma level of both risperidone and 9-OH-risperidone together ("active moiety") was 41.6 +/- 26.6 ng/ml. Significant improvements in PANSS scales and the various subscales ensued. There was a positive linear correlation between active moiety plasma levels and dose (r = 0.291, p = 0.015) and between risperidone and 9-OH-risperidone levels (r = 0.262; p = 0.016). Nonresponders to pharmacotherapy (PANSS-Improvement < 30%) showed significantly higher active moiety plasma levels (49.9 +/- 30.7 ng/ml) than responders (38.2 +/- 17.0 ng/ml; p = 0.045) without significantly higher oral doses (p = 0.601). Patients with longer illness duration (> or = 3 years) had significantly higher plasma drug levels than those with a shorter course (< 3 years; p = 0.039). Extrapyramidal side effects (EPS) and plasma levels were not correlated (r = 0.028; p = 0.843), but higher plasma levels at week 2 predicted an incidence for EPS (p < 0.050). Accordingly, patients initially receiving higher oral doses of risperidone were significantly more likely to respond with EPS in the trial course. Eight patients (9.8%) were heterozygous carriers of the CYP2D6 allele *4. CYP2D6 polymorphisms did not predict clinical response, but predicted a tendential increase in the plasma risperidone to 9-OH-risperidone ratio (0.5 +/- 0.6 vs. 1.9 +/- 1.8; p = 0.120). CONCLUSIONS The major finding was that responders to risperidone treatment had significantly lower blood levels of risperidone and 9-OH risperidone than patients who did not respond to the treatment despite administration of similar oral doses. The observed CYP2D6 polymorphisms did not contribute to altered clinical efficacy, but affected risperidone to 9-OH-risperidone ratios. Increased plasma levels of the active moiety in patients with longer illness may represent general aging effects. Conversely, the observed higher plasma levels in nonresponders may derive from unaccounted genetic metabolism abnormalities or Phase II metabolism disturbances. Patients initially receiving higher oral risperidone doses were more likely to respond with extrapyramidal side effects which reaffirms the need for careful titration. The high inter-individual variability in risperidone and 9-OH-risperidone metabolization and the relationship between clinical outcome and plasma levels warrants regular plasma level monitoring of both compounds to assess for the clinically relevant active moiety.