BACKGROUND We aimed to explore the risk factors that affect the serum concentration of sodium valproate (VPA-Na) in patients with epilepsy and to provide references for the rationale of the use of VPA-Na. MATERIAL AND METHODS The enzyme-multiplied immunoassay technique was used to determine the serum VPA-NA concentrations of 109 patients, and the results were retrospectively analyzed and summarized. A multivariate logistic regression model was used to analyze substandard serum VPA-Na concentrations. RESULTS Fifty-six patients (51.38%) treated with VPA-Na tablets were within the effective treatment range of 50-100 μg/mL, while 53 patients (48.62%) were out of the treatment range. The results indicated that the standard-reaching rate of serum drug concentration in the juvenile group was higher than that in the adult and elderly groups; the standard-reaching rates of serum drug concentrations in the low-dose group and the intermediate-dose group were lower than that in the high-dose group; and the standard-reaching rate of serum drug concentration in the group receiving carbapenems in combination was lower than that in the non-combination group; all differences were statistically significant. The combination with carbapenems and enzyme inducers was an independent risk factor for VPA-Na serum concentration below the target level in hospitalized patients. CONCLUSIONS To improve clinical efficacy and reduce the occurrence of adverse reactions, there is a need for therapeutic drug monitoring of VPA-Na. Moreover, individual administration should be implemented when VPA-Na tablets are used in the treatment of epilepsy because of the significant fluctuation in VPA-Na blood concentration.