To assess regional variation within England in the proportion of people with survey-defined hypertension who were on treatment, and hypothesize if this was due to chance or confounding. Data from three annual, cross-sectional health examination surveys, the Health Survey for England. Nationally representative random samples of the free-living general population were visited by an interviewer and a nurse. Blood pressure was measured with an automated monitor using a standardized protocol (2005: n=5321, 2006: n=10,213, 2007: n=4848). Hypertension was defined as systolic blood pressure ≥140mmHg, diastolic blood pressure ≥90mmHg, and/or taking prescribed medication to lower blood pressure. In London, a higher proportion of participants with survey-defined hypertension were on anti-hypertensive medication in each separate year's sample compared with the rest of England [2005-2007 average: 61% men, 66% women in London; 43% men, 55% women in England (P for London vs rest of England <0.001 for each sex)]. Regression analysis showed that this regional effect [odds ratio (OR) 1.47 95% confidence interval (CI) 1.94-2.47, P=0.031] was no longer significant after adjustment for demographic and socio-economic factors (OR 1.37, 95% CI 0.94-1.98, P=0.101), but was strengthened (OR 1.69, 95% CI 1.09-2.60, P=0.018) by including longstanding illness, diabetes, cardiovascular disease and health behaviours in the model. The proportion of hypertensive patients on anti-hypertensive medication was consistently above the national average in London, and this was associated with personal characteristics. Comorbidities increased the effect, even after adjustment for personal characteristics. This result may be due to greater population mobility in London, with more people having new patient health checks. Understanding this variation could enhance treatment nationally and internationally.