Context: Epidemiological studies have observed associations between diabetes and a number of different cancers. Yet the association with cancer overall and the interrelationship of diabetes and obesity with cancer have been unclear. Objective, Design, Setting, and Participants: We evaluated the association between self-reported diabetes and cancer incidence in the NIH-AARP (National Institutes of Health-American Association of Retired Persons) Diet and Health Study, a prospective cohort in which 295 276 men and 199 591 women completed a questionnaire in 1995-1996 and were followed up for cancer through 2006. Main Outcome Measures: Hazard ratios (HRs) and 95% confidence intervals (CIs) for cancer incidence, overall and by type, were estimated from multivariate Cox proportional hazards models. Results: Diabetes was positively associated with total incident cancer in women (1.07, 95% CI 1.02-1.12) but inversely in men (0.96, 95% CI 0.93-0.98). However, diabetes was inversely associated with prostate cancer (HR 0.74, 95% CI 0.70-0.78), which constituted 42% of cancers in men. After excluding prostate cancer, diabetes was also positively associated with cancer in men (HR 1.09, 95% CI 1.04-1.14). By site, diabetes was positively associated with anal, bladder, colon, kidney, liver, pancreatic, rectal, and stomach cancers and in women with endometrial cancer. We also evaluated the joint effect of obesity and diabetes and observed that diabetes conferred additional risk, beyond that of overweight or obesity, for cancer overall, excluding prostate, and for certain sites including the bladder, colon, endometrium, kidney, liver, pancreas, rectum, and stomach. Conclusion: Our results suggest an etiological role for diabetes in a number of cancers, independent of obesity, and that preventing diabetes may contribute to reduced cancer risk.