Contradictary results have been reported indicating both increased and reduced risks for malignancies in diabetic patients. This may possibly be due to difficulties in the clinical diagnosis of carcinomas and inaccuracies in the determination of diabetic conditions in the autopsy studies. Since glomerular microangiopathy is a typical feature of long-term diabetes, we performed a retrospective statistical analysis on 5000 consecutive, non-selected autopsy cases with particular reference to the presence/absence of microangiopathy in diabetic individuals. In our study group, we found a total incidence of 9.8% (n = 488) diabetic patients of which 213 (4.3%) had a histologically confirmed significant glomerulosclerosis and a total of 34% patients with verified carcinoma (n = 1699). The age- and sex ratios were matched between diabetic, non-diabetic and carcinoma patients. Systemic and coronary arteriosclerosis were significantly higher in diabetics than non-diabetics (p < 0.0001). Most interestingly, the rate of carcinomas in the diabetic group with nodular and diffuse glomerulosclerosis was 2.5- (p < 0.0001) and 1.9-fold (p < 0.0027), respectively, lower than in the non-diabetic group. In addition, the statistical evaluation showed in the glomerulosclerotic diabetic group significantly lower rates of metastasis. Our retrospective statistical analysis on an unselected series of autopsy cases thus provides evidence that diabetes mellitus with glomerulosclerosis is associated with a significantly lower frequency of carcinomas when compared to individuals without renal microangiopathy. Since TGF-beta is assumed to play a crucial role both in diabetes and carcinogenesis/tumor progression, our findings suggest an altered cell-matrix interaction in diabetes, possibly exerted by chronic TGF-beta overexpression.