A single observer reviewed 842 of the 917 known diabetic patients registered with 40 GPs in the Poole area. Fifty-nine per cent (493) of those reviewed submitted a timed overnight urine collection to measure albumin excretion rate (AER) and overnight albumin/creatinine ratio (ON-Alb/Creat); 43 samples were excluded because of urinary tract infection and/or proteinuria. A random urine sample was obtained in 607 diabetic patients to measure the random albumin/creatinine ratio (R-Alb/Creat); 68 specimens were excluded because of infection and/or proteinuria, and in a further 10 samples urinary creatinine was not measured. Stepwise multiple regression analyses found significant associations with the following variables: for AER, blood glucose (p = 0.001), smoking category (p = 0.002), sex (p = 0.034), and systolic blood pressure (p = 0.035); for R-Alb/Creat, blood glucose (p = 0.001), retinopathy (p = 0.004), systolic blood pressure (p = 0.004), diastolic blood pressure (p = 0.015), coronary artery disease (p = 0.02), sex (p = 0.034), and vibration sense (p = 0.038). Interestingly, glycosylated haemoglobin was not a significant determinant of albuminuria in either analysis.