The relation between diabetes and prostate cancer risk was investigated in a combined analysis of two hospital-based case-control studies conducted in Italy and Greece, between 1985 and 1997. Cases were 608 men with incident prostate cancer, and controls were 1008 men admitted to hospital for acute non-neoplastic diseases. No material association between diabetes and prostate cancer was observed, with a multivariate odds ratio (OR) of 1.07 (95% confidence interval (CI) 0.68-1.66). Compared with men without diabetes, there was a non-significant increased risk of prostate cancer in those diagnosed with diabetes within the last 5 years (OR 2.04), while the ORs were 0.96 and 0.78 respectively for a diagnosis of diabetes 5-9 years and > or = 10 years ago.