Fixed oral doses of clozapine produce up to 45-fold interindividual variability among its serum levels in patients with treatment-resistant schizophrenia. Although the relationship between serum clozapine level and its therapeutic response is uncertain, the presence of a therapeutic window and level-dependent adverse effects require the estimation of serum clozapine levels. As routine therapeutic drug monitoring of clozapine is not feasible in many clinical settings, identification of clinical predictors of serum clozapine levels is desirable. Hence, we aimed to evaluate the clinical variables associated with serum clozapine levels. We assessed the sociodemographic and clinical profiles, cognition, disability and psychopathology of 101 consecutive patients with treatment-resistant schizophrenia on a stable dose of clozapine, using standard assessment schedules. We determined their serum clozapine levels using high-performance liquid chromatography with ultraviolet detection. While employing multivariate robust regression models, oral clozapine dose (P<0.001), caffeine intake (P=0.04) and Valproate comedication (P=0.005) were associated with serum clozapine levels. Serum clozapine levels above 750 ng/ml increased the risk of seizures (odds ratio 5.15; P=0.03). Clinical variables are useful to model a dosing nomogram for serum clozapine levels. The importance of caffeine consumption and Valproate comedication should be considered during clozapine dose adjustments to enhance its therapeutic response and safety profile.