Although the importance of coronary microvascular dysfunction (CMD) has been emerging, reliable biomarkers for CMD remain to be developed. We examined the potential usefulness of plasma concentration of serotonin to diagnose CMD in patients with suspected angina and unobstructive coronary arteries. We enrolled 198 consecutive patients (M/F 116/82, 60.2 ± 13.3 years old) who underwent acetylcholine provocation test and measured plasma serotonin concentration. Coronary microvascular dysfunction was defined as myocardial lactate production without or prior to the occurrence of epicardial coronary spasm during acetylcholine provocation test. Although no statistical difference in plasma concentration of serotonin [median (inter-quartile range) nmol/L] was noted between the vasospastic angina (VSA) and non-VSA groups [6.8 (3.8, 10.9) vs. 5.1 (3.7, 8.4), P = 0.135], it was significantly higher in patients with CMD compared with those without it [7.7 (4.5, 14.2) vs. 5.6 (3.7, 9.3), P = 0.008]. Among the four groups classified according to the presence or absence of VSA and CMD, serotonin concentration was highest in the VSA with CMD group. Importantly, there was a positive correlation between plasma serotonin concentration and baseline thrombolysis in myocardial infarction frame count (P = 0.001), a marker of coronary vascular resistance. The classification and regression trees analysis showed that plasma serotonin concentration of 9.55 nmol/L was the first discriminator to stratify the risk for the presence of CMD. In multivariable analysis, serotonin concentration greater than the cut-off value had the largest odds ratio in the prediction of CMD [odds ratio (95% confidence interval) 2.63 (1.28-5.49), P = 0.009]. Plasma concentration of serotonin may be a novel biomarker for CMD in patients with angina and unobstructive coronary arteries.