Objective: This study was undertaken to determine the hemoglobin A1c (HbA1c) and modified glucose-ketone index (mGKI) in children on different types of ketogenic diet (KD) for treatment of drug-resistant epilepsy, with attempts to evaluate their relationships with components of diet regime and other biomarkers. Methods:We conducted a cross-sectional study in children with drug resistant epilepsy aged between 6 months and 18 years, who were on various types of KD therapies without any change in regime for at least 3 months. Parental interview, review of medical records, and a single measurement for blood ketone, HbA1c, and plasma carnitine were performed. mGKI was the ratio of an average plasma glucose estimated from HbA1c to blood βhydroxybutyrate level.Results: Thirty-four patients were recruited with a median blood ketone of 2.90 mmol•L -1 and median HbA1c of 4.55%. Those on classical KD (cKD) had higher blood ketone (p = .031) and lower HbA1c (p = .010) and mGKI (p = .021) than those receiving modified Atkins diet, although both shared similar percentages of calories from carbohydrate (p = .211). The cKD and medium-chain triglyceride (MCT) KD groups had similar HbA1c (p = .252) and mGKI (p = .510).Blood ketone (p = .045) and the percentage of calories from MCT (p = .037) were the two main independent variables, inversely correlating with HbA1c. Other than plasma acylcarnitine (p = .047), neither blood ketone (p = .188) nor HbA1c (p = .170) could predict seizure reduction reliably. Both plasma acylcarnitine ≥ 6 μmol•L -1 (p = .013) and mGKI ≤ 2.2 (p = .013) were significantly associated with good seizure control.Significance: HbA1c could potentially be useful for monitoring KD adherence or, indirectly, systemic ketosis in nondiabetic children on KD for drug-resistant epilepsy. Plasma acylcarnitine and mGKI could be important biomarkers in the management of KD therapy.