We have compared the chemical and clinical characteristics of an immunonephelometric assay (INA), two immunoturbidimetric assays (ITA) and two semi-quantitative methods with those of a solid-phase radioimmunoassay (RIA) for measurement of urinary albumin (UA) concentration in 136 diabetic patients. INA and RIA had similar accuracy, and provided comparable results. However, RIA has slightly greater sensitivity than INA, which is easier and faster. Good agreement was also found between RIA and the two ITA methods, although one of these overestimated RIA values in the low-medium range (5-30 mg/l) of urinary albumin. ITA seems suitable for initial screening of albuminuria in diabetic patients but more sensitive procedures (such as RIA and INA) seem preferable for measurement of UA concentrations in the normal range. The two semi-quantitative methods showed high sensitivity but poor specificity, because of the large number of false positive results. About 50% of diabetic patients "positive" by these methods did not have microalbuminuria. The utility of these methods is questionable, because many samples from diabetic patients need to be reassayed by a more specific and sensitive assay such as the RIA, INA or ITA methods.