AIMS: To estimate the size of regression to the mean with ambulatory blood pressure (ABP) measurement. METHODS: Participants from five studies who had repeated blood pressure (BP) measurements using office and ambulatory devices were included. Regression to the mean was calculated following participants being grouped by baseline BP categories. Regression dilution ratio was calculated for groups defined by each baseline BP variable. RESULTS: High baseline ABP readings were substantially lower on long-term follow-up, and low baseline readings tended to be higher. Regression to the mean was observed for all ABP parameters; for systolic and diastolic measures; and for intervention and control groups. For example, among those with baseline 24-h SBP of at least 150 mmHg, mean baseline and follow-up BP was 156 and 141 mmHg, respectively; whereas those with baseline 24-h SBP of less than 120 mmHg, mean baseline and follow-up BP was 113 and 119 mmHg, respectively. Regression to the mean was the greatest for night-time ABP. Regression dilution ratios calculated from control groups were 0.52, 0.53, 0.38 and 0.60 for 24-h, daytime, night-time and office SBP, respectively. Similar results were seen for diastolic measures. CONCLUSION: ABP is subject to considerable regression to the mean, which has implications for diagnosis and practise; for example, after initiating treatment for hypertension some of the fall in ABP will be because of regression to the mean. Furthermore, associations of ABP with cardiovascular disease will be substantially underestimated if analyses are not adjusted for regression to the mean, especially for night-time ABP. Replication studies are needed to confirm these findings.