To assess the validity of urine albumin concentration (UAC) and the urine albumin:creatine ratio (UACR) in a random urine specimen (RUS) for screening diabetic nephropathy in Korea, a total of 105 ambulatory diabetes mellitus patients (male:female, 52:53), ages 40-75 years (median 59 years) collected 105 RUSs after completing a timed 24 hour urine collection. Albumin was measured by immunonephelometry. According to the timed urinary albumin excretion rate (UAER) measured in the 24 hour collection (criterion standard), samples were classified normoalbuminuric (UAER < 20 micrograms/min; n = 50), microalbuminuric (UAER 20-200 micrograms/min; n = 30), and macroalbuminuric (UAER > 200 micrograms/min; n = 25). The receiver operating characteristics (ROC) curve of UAC and UACR in a RUS for screening of microalbuminuria (normo- and microalbuminuric samples; n = 80) and macroalbuminuria (micro- and macroalbuminuric samples; n = 55) were plotted. Pearson's coefficients of correlation of 24 hour UAER vs. UAC and UACR were 0.81 and 0.75, respectively (P < 0.001). The point of intersection with a 100%-to-100% diagonal for microalbuminuria were as follows: 31.0 mg/l for UAC and 32.5 mg/g for UACR; for macroalbuminuria 181 mg/l for UAC and 287.3 mg/g for UACR. The sensitivity and specificity of the cut-off points for microalbuminuria were 77% and 82% for UAC and 77% and 92% for UACR. The sensitivity and specificity of the cut-off points for macroalbuminuria were 84% and 90% for UAC and 88% and 90% for UACR. In present study, no difference was observed when comparing the performance of UAC and UACR based on a statistical comparison by McNemar test. The repeated measurements of UAC and UACR in the same individual were statistically similar and were correlated with each other. Based on these results, albumin measurements (UAC and UACR) in a RUS were considered as a valid test for screening diabetic nephropathy.