Proteinuria in patients with glomerular disease has a circadian rhythm, but for creatinine such a rhythm is either absent or of low amplitude. We found in 18 of 23 admitted patients (group I) and in seven outpatients (group II) a marked circadian rhythm of the protein: creatinine ratio. Estimates of 24-h proteinuria were obtained by multiplying the protein:creatinine ratio of 3-h urine samples with 24-h creatinine excretion, calculated from age, sex and bodyweight. Estimated proteinuria could be as low as 19% or as high as 349% of actually measured 24-h proteinuria; the mean SD was 23%. The best estimate was obtained with the 06.00-09.00 hours urine samples. The estimates correlated better with actually measured 24-h proteinuria than the protein: creatinine ratio per se correlated with the 24-h proteinuria. Day-to-day variation of proteinuria estimates from samples taken at the same time of the day was of similar magnitude as day-to-day variation of actual 24-h proteinuria. We conclude that the usefulness of the protein: creatinine ratio of a random urine sample for estimation of proteinuria is limited, because of the circadian rhythm of proteinuria. However, samples collected at a fixed time of the day are an acceptable alternative for 24-h urine collections in the clinical follow-up of individual patients.