The aim of the present study, which is part of a larger population survey, is to provide an Italian estimate of the total prevalence of diabetes taking into account the fraction of known diabetes among the non-responders. The study consists of a population survey. Out of the population aged over 40 ys. of two municipalities, Casalbuttano and Vescovato, in the Health District of Cremona, a random sample of 2,229 subjects was selected to undergo an OGTT. In addition, a series of anamnestic, clinical, and laboratory data were collected for each individual. Overall response rate was 63%. Prevalence of known diabetes was 7.8% (6.5%-9.1%, 95% CI), (F = 7.9%, M = 7.6%). The comparison between responders and the telephone-interviewed non-responders showed an unpredictable and age- and sex-dependently varying frequency of known diabetes, even if the overall differences were not statistically significant. This suggests that, to avoid biases, many efforts should be spent to collect information about diabetes also for non-responders. Prevalence of unknown diabetes was 3.2% (2.2%-4.2%, 95% CI), (F = 4.0%, M = 1.8%) and total diabetes was 10.9% (9.3%-12.5%, 95% CI) (F = 11.9%, M = 9.4%). In contrast to other countries, the relationship between unknown and known diabetes was less than 1 to 2, and the proportion of unknown diabetes was almost negligible below the age of 70. A positive family history and a lower level of triglycerides were associated with awareness of the disease (p = 0.0005 and p = 0.006, respectively), thus suggesting different behaviour towards one's own health between known and unknown diabetic subjects.