Blood pressure measurements made with a Dinamap oscillometric blood pressure recorder by two observers on a total of 2596 male and female subjects aged between 25 and 59 living in nine British towns were examined for evidence of observer differences. After adjusting for age, body mass index and town, significant differences between the two observers were found for systolic pressure, 3.07 mmHg (P = 0.001) for male subjects and 2.08 mmHg (P = 0.036) for female subjects. Observer differences for mean arterial pressure and diastolic pressure were less than 1 mmHg and were not significant. The magnitude of observer differences also varied between towns and, when tested statistically, was significant in male subjects for systolic (P = 0.011), mean arterial (P = 0.009) and diastolic pressure (P = 0.002); for female subjects it was significant only for diastolic pressure (P = 0.023). We conclude that although the highly automated Dinamap has kept observer variation within acceptably low limits in field study conditions, the problem has not been altogether eliminated.