Pharmacokinetic parameters were compared between 2 dosages of valproic acid (VPA) sustained-release (SR) formulation in psychiatric patients. A total of 66 psychiatric patients (21 women and 45 men; age range, 18-60 years) receiving 500 mg/d (n = 28) or 1000 mg/d (n = 38) of VPA SR for at least 4 weeks were recruited into the study. A 5-mL blood sample was collected into a plain tube and immediately centrifuged for plasma. Separation of free VPA was further done using Centrifree micropartition devices. Both total and free VPA concentrations (C(total) and C(free)) were analyzed by TDx analyzer with fluorescence polarization immunoassay technique. The patients' characteristics and pharmacokinetic parameters were compared between the 2 dosage groups using independent t test or χ² test where appropriate. The results show that the increment in C(total) (mg/mL) for every milligram-per-kilogram increment in dosage was decreased from 7 ± 3 to 4 ± 1 (mg/L) / (mg/kg) when the total VPA clearance (CL(total)) increased from 6.1 ± 2.6 to 9.0 ± 3.1 mL/kg per hour with the increasing dose in the 500 mg/d and 1000 mg/d groups, respectively (P < 0.05). The increment in C(free) [0.6 ± 0.3 vs 0.5 ± 0.2 (mg/L) / (mg/kg)] and CL(free) (80.4 ± 41.5 vs 92.2 ± 47.6 ml/kg per hour) were not significantly different. Owing to the saturation of protein binding, percent free VPA was significantly increased from 8 ± 3 to 11 ± 3 when the dose was increased from 500 to 1000 mg/d (P < 0.05). In conclusion, an increase in the VPA dose resulted in a disproportional increase between dosage and C(total), whereas a proportional increase between dosage and C(free) still existed. Therefore, our study suggests that the therapeutic range of C(free) is 4 to 12 mg/L based on the therapeutic range of C(total) (45-100 mg/L) for general psychiatric conditions.