PURPOSE: Coffee is a widely consumed beverage and small health effects of substances in coffee may have large public health consequences. It has been suggested that caffeine in coffee increases the risk of hypertension. We performed a meta-analysis of randomized controlled trials of coffee or caffeine and blood pressure (BP). DATA IDENTIFICATION: BP trials of coffee or caffeine published between January 1966 and January 2003 were identified through literature databases and manual search. STUDY SELECTION: A total of 16 studies with a randomized, controlled design and at least 7 days of intervention was selected, comprising 25 strata and 1010 subjects. DATA EXTRACTION: Two persons independently obtained data on sample size, type and duration of intervention, changes in BP and heart rate (HR), and subjects' characteristics for each trial. Meta-analysis was performed using a random-effects model. RESULTS: A significant rise of 2.04 mmHg [95% confidence interval (CI), 1.10-2.99] in systolic BP and 0.73 mmHg (95% CI, 0.14-1.31) in diastolic BP was found after pooling of coffee and caffeine trials. When coffee trials (n = 18, median intake: 725 ml/day) and caffeine trials (n = 7, median dose: 410 mg/day) were analysed separately, BP elevations appeared to be larger for caffeine [systolic: 4.16 mmHg (2.13-6.20); diastolic: 2.41 mmHg (0.98-3.84)] than for coffee [systolic: 1.22 mmHg (0.52-1.92) and diastolic: 0.49 mmHg (-0.06-1.04)]. Effects on HR were negligible. CONCLUSIONS: Regular caffeine intake increases BP. When ingested through coffee, however, the blood pressure effect of caffeine is small.