BACKGROUND The aim of this naturalistic exploratory study was to examine whether blood antipsychotic drug concentrations can predict rehospitalizations in chronically medicated patients. METHODS The study included schizophrenic outpatients under clozapine (CLZ) maintenance treatment, supervised by therapeutic drug monitoring (TDM). Patients were observed for a period of 21 months. Their on average monthly measured plasma levels and the date of rehospitalizations were recorded. The variability of the first 3 CLZ plasma levels, measured in 3.6 months, was compared between patients with and without rehospitalization. RESULTS 23 patients participated of which 6 patients were rehospitalized. Mean plasma concentrations of CLZ were similar in patients without (471 ± 180 ng/mL) and with rehospitalization (446 ± 266 ng/mL). However; coefficients of variation (CV) of plasma concentrations in the first 3 blood samples differed significantly between the rehospitalized and non-rehospitalized groups (37.1% vs. 13.0%, respectively, P = 0.012). A CV ≥ 19.8% was predictive for later rehospitalization with 100% sensitivity and 70.6% specificity. CONCLUSIONS Variability in CLZ plasma concentrations may be useful in identifying patients at risk of relapsing under maintenance therapy. Because of the small number of patients the findings need to be confirmed in a larger study.