BACKGROUND: Approximately 30 % of patients with treatment-resistant depression (TRD) respond to ketamine; however, no replicable predictors of response have been reported. The imbalance between excitatory and inhibitory neurotransmissions may be implicated in the mechanism of action of ketamine. This study aimed to evaluate whether the ratio of glutamate and glutamine (Glx) to GABA levels at baseline in the dorsal anterior cingulate cortex (dACC) could predict ketamine response in patients with TRD. METHOD: This exploratory study analyzed data from a double-blind randomized clinical trial with an open-label extension study (jRCTs031210124). Fifteen participants in the ketamine group and 15 of 16 participants in the placebo group received repeated intravenous ketamine during the double-blind and open-label extension periods, respectively. We measured Glx and GABA levels in the dACC before and after treatment during the double-blind period using proton magnetic resonance spectroscopy. The 17-item Hamilton Depression Rating Scale (HDRS-17) was measured for depressive symptomatology. General linear models were used to examine the relationship between baseline Glx/GABA ratio and HDRS-17 score changes. RESULT: Changes in HDRS-17 scores (mean (±SD)) following ketamine treatment were -4.9 (6.5) and -4.9 (5.2) in the double-blind and open-label periods, respectively. A higher baseline dACC Glx/GABA ratio was correlated with greater improvement in HDRS-17 (β = -0.42, p = 0.040). In the ketamine group, a reduction in the dACC Glx/GABA ratio was correlated with greater HDRS-17 improvement (β = 0.74, p = 0.009) with no such association in the placebo group. CONCLUSION: These results suggest that excitatory-inhibitory imbalance in the dACC may predict the efficacy of ketamine in TRD.