In noninsulin dependent diabetes, elevation of urinary albumin excretion above the normal range is thought to predict increased mortality. A random urinary albumin/creatinine ratio and the following clinical characteristics were recorded on 405 patients with noninsulin dependent diabetes attending the Wellington Hospital diabetes clinic; (1) ethnic origin, (2) age, (3) sex, (4) age at diagnosis, (5) duration since diagnosis of diabetes, (6) body mass index, (7) systolic blood pressure, (8) diastolic blood pressure, (9) retinopathy, (10) insulin therapy, (11) antihypertensive therapy, and (12) the presence or absence of treated cardiovascular disease. The overall prevalence of elevated urinary albumin/creatinine ratio (greater than 2.2 mg/mmol) was 45%. Stepwise regression analysis showed that 12% of the variability of urinary albumin/creatinine ratio in our clinic population could be accounted for by ethnic origin of patients, with noncaucasians (the Maori, Pacific Islanders and Asians) having a higher urinary albumin excretion than caucasians (p less than 0.001). Systolic blood pressure, retinopathy and cardiovascular disease predicted a further 8% of this variability. The cause of this ethnic difference in urinary albumin excretion and its possible relationship to excess mortality in noncaucasians is not yet known.