The procedures for obtaining blood pressure measurements, and their quality of performance, should be similar between epidemiological studies to allow a valid comparison of blood pressure distribution and prevalence of hypertension between study populations. However, considerable methodological variation currently exists as there are several international guidelines on blood pressure measurement and a standard method of measurement has not yet been agreed. This paper reviews the literature in order quantitatively to define systematic differences between blood pressure measurements resulting from differences in arm used, body position, number of measurements, verbal communication, cuff size applied and measuring device used. The results were applied to differences in procedure between European Prospective Investigation into Cancer and Nutrition (EPIC) study centres, showing that blood pressure measurements may systematically differ between centres by as much as 10 mmHg for procedural reasons alone. For most reviewed methodological differences, it was possible to at least estimate a range of systematic error. There are, however, additional possible causes of systematic variation, for example arm position, postprandial hypotension and diurnal blood pressure variation, that could not be taken into account. It may be possible to make only a partial correction for the systematic inter-centre variation of the EPIC blood pressure measurements. This study highlights the importance of standardized protocols for blood pressure measurements in epidemiological studies.