Aims: Coronary vascular dysfunction is linked with poor cardiovascular prognosis in patients without obstructive coronary artery disease (CAD) but a critical appraisal of the literature is lacking. Methods and results: We performed a systematic review and meta-analysis to quantify the cardiovascular risk associated with endothelial dependent and non-endothelial dependent coronary vascular dysfunction in patients with normal or non-obstructive CAD (epicardial stenosis < 50%). Prospective cohort studies that reported coronary vascular dysfunction at baseline and cardiovascular outcomes at follow-up were included. We identified 52 papers of which 26 were included in the meta-analyses. Study populations included stable angina (n = 15), heart failure (n = 4), diabetes (n = 2), hypertrophic obstructive cardiomyopathy (n = 2), chronic kidney disease, aortic stenosis and left atrial enlargement (each n = 1): RR estimates were similar in patients with stable angina and other patient groups. For epicardial endothelial dependent dysfunction (six studies, 243 events in 1192 patients) the summarized RR was 2.38 (95% confidence intervals (95% CI) 1.74–3.25), for non-endothelial dependent dysfunction assessed as coronary flow velocity reserve (CFVR) by echocardiography (10 studies, 428 events in 5134 patients) RR was 4.58 (95% CI 3.58–5.87) and for coronary flow reserve (CFR) by PET (10 studies, 538 events in 3687 patients) RR was 2.44 (95% CI 1.80–3.30). However, RR estimates were robust in a series of sensitivity analyses. Conclusion: The presence of coronary vascular dysfunction in patients with normal or non-obstructive CAD predicts adverse cardiovascular outcome. Multicentre studies and uniform guidelines for assessing coronary vascular dysfunction are encouraged.