The total prevalent population of insulin-treated diabetic subjects (1499 patients) in Fyn County, Denmark, was followed for 8 1/2 years (beginning 1 July 1973). All cases of cancer diagnosed during this period were identified by a record cross-check with the Danish Cancer Registry. In total, 45 and 37 cancer cases were identified among male and female patients, respectively, corresponding with observed/expected ratios of 1.37 (p = 0.03) and 1.08 (p = 0.65), respectively. Specified for site, a significant excess of cancer of the pancreas was found (six cases observed vs 2.4 expected, p = 0.02), but the excess was reduced by excluding cases where the cancer patient had diabetes as an early sign. Overall, no indications were found of an association between diabetes and cancer, whether in general or regarding particular cancer types.