In a study to monitor the acceptability of using a timed overnight urine collection to measure microalbuminuria in a large diabetic population, it was found that only 184 (69.4%) of 265 diabetics returned the urine collection to the hospital laboratory. Three proposals for an initial screening test to predict high risk microalbuminuria were analysed. Proposal I: a random urinary albumin concentration (RAC) greater than 25 micrograms/ml predicts an overnight albumin excretion rate (AER) greater than 30 micrograms/min. In 159 diabetics, the sensitivity of this screening test was 56%, specificity 81%, and predictive value 15%. Proposal II: an albumin concentration greater than 20 micrograms/ml in an early morning urine sample predicts an overnight AER greater than 30 micrograms/min. In 175 diabetics, the sensitivity of this screening test was 86%, specificity 97%, and predictive value 71%. Proposal III: an albumin/creatinine ratio (albumin mg/l creatinine mmol/l) greater than 3.5 in an early morning urine sample predicts an overnight AER greater than 30 micrograms/minute. In 171 diabetics, the sensitivity of this screening test was 100%, specificity 95%, and predictive value 64%. It is concluded that an albumin/creatinine ratio performed on an early morning urine sample brought to the diabetic clinic would be a useful initial screening test to detect an overnight AER greater than 30 micrograms/min.