This study examined the prevalence of diagnosed diabetes mellitus in a defined population over 13 years by undertaking cross-sectional surveys on 3 occasions between 1983 and 1996. The study population consisted of all the people registered with 10 general (primary care) practices at the time of each survey; 90660 in 1983/4; 97122 in 1988/9; and 86287 in 1996. Ascertainment of cases was by a surveillance programme in general practice and the hospital diabetes department. The number of diabetic patients increased significantly over the study period: in 1983/4, there were 917 patients, crude prevalence 1.01% (95% CI 0.95-1.08%); in 1988/9, 1150 patients, crude prevalence 1.17% (1.12-1.25%); and in 1996, 1604 patients, crude prevalence 1.86% (1.77-1.95%). The prevalence adjusted to the age and sex distribution of the UK was 0.97% (95% CI 0.90-1.03%) in 1983/4, 1.05% (0.99-1.11%) in 1988/9 and 1.55% (1.48-1.63%) in 1996. The main increase in prevalence was due to Type 2 diabetes mellitus, crude prevalence 0.75% (95% CI 0.69-0.81%) in 1983/4, 0.92% (0.86-0.98%) in 1988/9 and 1.52% (1.44-1.60%) in 1996 rather than Type 1 diabetes mellitus, crude prevalence 0.25% (0.21-0.28%) in 1983/4, 0.25% (0.22-0.28%) in 1988/9 and 0.34% (0.30-0.38%) in 1996. During the study period, the crude prevalence of diagnosed diabetes was significantly greater in men than women; in 1983/4 men 1.1% (95% CI 1.00-1.20%) versus women 0.93% (0.84-1.02%); in 1988/9, men 1.31% (1.21-1.41%) versus women 1.07% (0.98-1.16%); and in 1996, men 2.13% (2.00-2.27%) versus women 1.60% (1.49-1.72%). This difference was statistically significant in the 1988/9 and 1996 surveys. In conclusion, over 13 years there was a significant increase of 83.6% in the prevalence of diagnosed diabetes mellitus in the Poole area, with the UK age and sex adjusted prevalence increasing by 60.7%.