BACKGROUND Numerous surveys carried out in India report the high prevalence of type 2 diabetes. Such studies have not included the population of Kerala. We estimated the prevalence of type 2 diabetes mellitus in Neyyattinkara taluk, Thiruvananthapuram district, Kerala state. METHODS All panchayat wards in the taluk were grouped into urban, highland, midland or coastal, and one ward from each stratum was randomly selected for the study. All households were listed and adults, 20 years or older, screened for high (> 110 mg/dl) random blood sugar (RBS) by a glucometer test. Those with high RBS were reassessed by a fasting oral glucose tolerance test (OGTT), consisting of initial examination of venous blood for fasting plasma glucose (FPG) values, administration of 75 g of glucose dissolved in distilled water, and examination of venous blood for postprandial plasma glucose (PPPG) exactly 2 hours after the administration of glucose. Diabetes was diagnosed according to the World Health Organization criteria as either FPG > 139 mg/dl, or PPPG > 199 mg/dl, or both. Impaired glucose tolerance (IGT) was diagnosed if PPPG was 140-199 mg/dl. RESULTS Out of 4988 eligible subjects, 3899 were available for the study, a response rate of 78.2%. Response was highest in the highland area (86.2%), and lowest in the coastal area (73.6%). The overall crude prevalence rate of type 2 diabetes was 5.9%. It was highest in the urban (12.4%), followed by midland (8.1%), highland (5.8%), and coastal (2.5%) regions. Ageing was associated with greater prevalence of type 2 diabetes in all regions and both sexes. Women showed a higher prevalence in the highland and coastal areas and men in the urban and midland areas. When compared to a population with standard age structure suggested by the World Health Organization for international comparisons, prevalence in the age group 30-64 years was found to be 16.9% in the urban, 10.1% in the midland, 6.8% in the highland and 3.6% in the coastal regions, respectively. Overall age-adjusted prevalence of type 2 diabetes in 30-64-year-olds in Neyyattinkara was 9.2% among men, 7.4% among women, and 8.2% for all persons. There was not much difference in prevalence if the American Diabetes Association criteria for diagnosis were used. The prevalence of impaired glucose tolerance was fairly low in this population. Out of 229 diabetics in the sample, 175 (76.5%) were already diagnosed and under treatment, while our survey identified 54 new diabetics (23.5%). CONCLUSIONS Though prevalence of type 2 diabetes is high in this population, the detection rate is also high. However, impaired fasting glucose and impaired glucose tolerance are low. The reasons for this need to be elucidated.