The hypothesis that the prevalence of cardiovascular risk factors in people with diabetes is inversely related to socio-economic status was tested. Demographic and biochemical data were collected on 1246 patients, aged 20-69 years, attending a hospital diabetes clinic. This is estimated to represent between 71% and 78% of all people of this age with a diagnosis of diabetes in the health authority. In total, 296 people were classified as Type 1 (insulin-dependent) diabetic patients (age of onset < 31, now on insulin). Using data from the 1991 census a deprivation score was ascribed to each individual according to their area (enumeration district) of residence. The total study population was ranked by deprivation score and divided into quintiles. The relationships between means and quintiles of deprivation were assessed by ANOVA for linear trend, and between proportions and quintiles of deprivation by the chi-squared test for trend. In Type 1 diabetes increasing quintiles of deprivation were significantly related to mean serum cholesterol (p < 0.01) and proportion smoking (p < 0.01), and in Type 2 (non-insulin-dependent) diabetes to mean body mass index (p < 0.001), proportion smoking (p < 0.001), and proportion with proteinuria (p < 0.05). The need for health measures to prevent cardiovascular disease in people with diabetes is greatest in deprived areas.