To provide complete prevalence data on diabetes mellitus in Italy (diagnosed and undiagnosed), a population survey was performed in the Health District of Cremona, a representative area of the Po river (north Italy). The survey is characterised by particular attention being paid to methodology, non-responders being investigated for possible selection biases affecting diagnosed and undiagnosed diabetes prevalence estimations. Out of a population aged 44 years or older from three municipalities, a random sample of 3097 subjects was selected to undergo an oral glucose tolerance test. In addition, past medical history, clinical and laboratory data were collected. A total of 1797 subjects participated (58%), and information on known diabetes status was obtained for 2618 persons (85%), also including 826 interviewed non-participating subjects. Overall rates were age-standardised according to the 1991 Italian census. Overall prevalence and 95% confidence interval of diagnosed diabetes was 8.5% (6.9-10.1) in males and 7.9% (6.7-9.3) in females over the age of 44 years; previously undiagnosed diabetes was 2.5% (1.4-3.6) in males and 3.4% (2.1-4.7) in females; glucose intolerance was 7.7% (5.7-9-7) in males and 8.9% (7.0-10.8) in females. Total diabetes prevalence above age 44 years, developed-world age, and sex standardised, was 10.7%. This study provides the first reliable prevalence estimation of impaired glucose tolerance, diagnosed and undiagnosed diabetes in Italy, according to World Health Organization criteria, and one of the few figures for Southern Europe. The role of body mass index on both prevalence of diabetes and cluster of cardiovascular risk factors is considered, with implications for prevention.