Mildly increased urinary albumin excretion rates and concentrations, below the quantity normally detected by conventional urinary protein and albumin methods, have prognostic significance for the development of nephropathy in patients with diabetes mellitus. The authors evaluated the automated Behring Nephelometer using Behring reagents for the detection of low level urinary albumin. Within run coefficients of variation (CVs, N = 20) are 1.7%, 1.3%, and 2.4% at mean urinary albumin levels of 16, 70, and 217 mg/L, respectively. Between run CVs (N = 20) are 4.5%, 2.6%, and 4.4% at mean albumin levels of 19, 71, and 239 mg/L, respectively. The method is sensitive to 3 mg/L. Hemoglobin, immunoglobulins, bilirubin, urea, and radiographic contrast media beyond a few hours of injection show no significant interference at levels normally expected from clinical specimens. Analysis is unaffected by pH within the physiologic range. Most urine specimens are stable for at least eight days when refrigerated at 4 degrees C. Specimen centrifugation before analysis is essential to avoid a negative bias that occurs when analyzing uncentrifuged refrigerated samples. Preanalytical freezing produces results higher than those observed in fresh or refrigerated samples. The authors conclude that automated nephelometry using the Behring Nephelometer is a convenient, simple, and accurate technique for the determination of low level urinary albumin.