It is not clear whether 24-hour or overnight urine collections should be used to identify patients with incipient diabetic nephropathy (defined as persistent urinary albumin excretion rate [AER] of 20 to 200 micrograms/min). We therefore studied diurnal variations in AER in type I diabetics with normal AER (n = 16) and incipient (n = 12) and clinical (defined as persistent AER greater than 200 micrograms/min) nephropathy (n = 12), and in healthy controls (n = 24). In all groups AER was lowest at night. In some patients of all groups, marked diurnal variations were observed. Twenty-four-hour urine collections classified all patients correctly. Overnight urine collections, however, misclassified patients with incipient nephropathy as having normal AER in 4 of 12, 7 of 12, or 3 of 7 cases, depending on which cutoff level was used. We conclude that 24-hour urine collections are more sensitive than overnight samples in identifying patients with incipient diabetic nephropathy.