Background: Valproic acid (VPA) is a first-line anti-epileptic drug with broad efficacy. Due to significant individual differences in its metabolism, therapeutic drug monitoring (TDM) is commonly used. However, the recommended therapeutic range (50-100μg/mL) is inadequate for predicting clinical outcomes. Additionally, relationship between VPA metabolites and clinical outcomes remains unclear. Methods: In this retrospective study, 485 Chinese Southern Han epilepsy patients receiving VPA monotherapy were analyzed after reaching steady-state levels. Plasma concentrations of VPA and its five main metabolites were determined by LC-MS. We assessed the relevance of the recommended therapeutic VPA range for clinical outcomes and explored the association between VPA/metabolite levels and treatment efficacy/adverse effects. In vitro experiments were conducted to assess 4-ene-VPA hepatotoxicity. Results: The therapeutic range of VPA exhibited no significant correlation with clinical outcomes, and plasma concentrations of VPA failed to serve as predictive indicators for treatment response/adverse effects. Treatment responders had higher 2-PGA concentrations (median 26.39 ng/mL vs. 13.68 ng/mL) with a threshold of 36.5 ng/mL for optimal epilepsy treatment. Patients with abnormal liver function had a higher 4-ene-VPA median concentration (6.41μg/mL vs. 4.83μg/mL), and ratio of 4-ene-VPA to VPA better predicted VPA-induced hepatotoxicity (AUC=0.718) than 4-ene-VPA concentration. In vitro experiments revealed that 4-ene-VPA was more hepatotoxic than VPA in HepaRG and L02 cell lines. Conclusions: Total plasma VPA concentration does not serve as a predictor for clinical outcomes. 2-PGA concentrations may be associated with efficacy, while the ratio of 4ene-VPA to VPA may be considered as a better biomarker (threshold 10.03%) for VPAinduced hepatotoxicity.