The aim of the study was to define the relationship between the presence of proliferative diabetic retinopathy and nephropathy with objectively defined autonomic neuropathy in non-insulin-dependent diabetes mellitus (NIDDM) patients. The research design used was a cross-sectional, case control study. A cohort of NIDDM patients was classified, according to five cardiovascular autonomic tests described by Ewing, as: (1) no involvement--no abnormal tests (n = 17); (2) cardiovascular autonomic neuropathy--two out of five abnormal tests (n = 18). Age, age at diagnosis, plasma creatinine, fasting plasma glucose, glycated haemoglobin and blood pressure measurements were not statistically different among the two groups. According to indirect ophthalmoscopy and the presence of macroproteinuria and microalbuminuria, respectively, patients were also classified as having proliferative, non-proliferative or no retinopathy and with or without nephropathy. The results showed a striking relationship between cardiovascular autonomic neuropathy and proliferative diabetic retinopathy. Relative odds for nephropathy, non-proliferative diabetic retinopathy and proliferative retinopathy were, respectively, 16.0, 10.1 and 34.7. When odds ratios were adjusted for the presence of nephropathy, hypertension, non-proliferative and proliferative retinopathy, only proliferative retinopathy was significantly associated (odds ratio, 7.1). It was concluded that in NIDDM the presence of cardiovascular autonomic neuropathy is strongly associated with proliferative retinopathy. Long-term prospective studies on large cohorts of patients must be done to evaluate if having autonomic dysfunction would be a risk factor or a risk indicator of an etiologic process underlying the development of proliferative retinopathy.