Associations between diabetes and multisite cancer risk were examined in a case-control study of 11,672 incident cancer cases (5341 men and 6331 women) and 47,768 cancer-free controls (14,199 men and 33,569 women). Using an unconditional logistic regression model and controlling for potential confounding factors, a past/present history of diabetes was associated with cancer risks for all sites in both men and women (odds ratios=1.44 and 1.39, 95% confidence intervals=1.28-1.62 and 1.19-1.62), respectively. Significantly increased risk was found for cancers of the pharynx, esophagus, colorectum, liver, pancreas, larynx and lung among men, and of the stomach, liver, lung and cervix uteri among women. Among participants with a family history of diabetes, a past/present history of diabetes was also associated with cancer risks of all sites among men and women (odds ratios=1.41 and 1.96, 95% confidence intervals=1.04-1.93 and 1.44-2.68, respectively). The associations were highest for pancreatic cancer among men and the stomach, colorectum and corpus uteri among women. Our findings support the conclusion that diabetes is associated with the etiology of cancer development, and we suggest that a family history of diabetes should be considered as a risk factor for several cancer sites among participants with a past/present history of diabetes.