RATIONALE: (R,S)-ketamine is a rapid and effective antidepressant drug that produces a response in two thirds of patients with treatment-resistant depression (TRD). The underlying biochemical differences between a (R,S)-ketamine responder (KET-R) and non-responder (KET-NR) have not been definitively identified but may involve serine metabolism. OBJECTIVES: The aim of the study was to examine the relationship between baseline plasma concentrations of D-serine and its precursor L-serine and antidepressant response to (R,S)-ketamine in TRD patients. METHODS: Plasma samples were obtained from 21 TRD patients at baseline, 60 min before initiation of the (R,S)-ketamine infusion. Patients were classified as KET-Rs (n = 8) or KET-NRs (n = 13) based upon the difference in Montgomery-Åsberg Depression Rating Scale (MADRS) scores at baseline and 230 min after infusion, with response defined as a ≥50 % decrease in MADRS score. The plasma concentrations of D-serine and L-serine were determined using liquid chromatography-mass spectrometry. RESULTS: Baseline D-serine plasma concentrations were significantly lower in KET-Rs (3.02 ± 0.21 μM) than in KET-NRs (4.68 ± 0.81 μM), p < 0.001. A significant relationship between baseline D-serine plasma concentrations and percent change in MADRS at 230 min was determined using a Pearson correlation, r = 0.77, p < 0.001, with baseline D-serine explaining 60 % of the variance in (R,S)-ketamine response. The baseline concentrations of L-serine (L-Ser) in KET-Rs were also significantly lower than those measured in KET-NRs (66.2 ± 9.6 μM vs 242.9 ± 5.6 μM, respectively; p < 0.0001). CONCLUSIONS: The results demonstrate that the baseline D-serine plasma concentrations were significantly lower in KET-Rs than in KET-NRs and suggest that this variable can be used to predict an antidepressant response following (R,S)-ketamine administration.