The present study examined whether levels of cadmium, and alphal- and beta2-microglobulin in urine (Cd-U, ac-MG-U, and beta2-MG-U, respectively) were reproducible in urine samples collected from the same subjects on multiple occasions. For this purpose, two databases on background exposure to cadmium in Japan-one from study I between 2000 and 2001 and the other from study II in 2002-were revisited to find 231 apparently healthy, nonpregnant, nonlactating adult women who participated in both studies and thus had provided two urine samples. The databases contained information on Cd-U, alphal,-MG-U, and beta2-MG-U, creatinine (CR), and specific gravity (SG) as well as smoking and other lifestyle factors. Of the 231 women, 195 who had never smoked were selected for the present analysis. Cd-U as well as alpha1-MG-U were reproducible (e.g., with correlation coefficients [r] between study I and II results of 0.4 to 0.6) when measured on two occasions 9 to 10 months apart. The r values were lower for beta2-MG-U (r0.3). Exclusion of urine samples with inadequate urine density(i.e., CR <0.5 or >3.0 g/L or SG <1.010 or >1.030) resulted in substantial improvement of the agreements between the two measures (e.g., r = 0.6 to 0.7 for Cd-U and alpha1-MG-U). CR and SG correlated closely with each other, especially in low-density urine samples (r >0.9), and therefore the effects of CR and SG could not be evaluated separately. In the overall evaluation,single determination (i.e., without repeated urine sampling) of Cd-U and alpha1-MG-U should be acceptable, and it may also be acceptable for beta2-MG-U. Use of samples with adequate urine density rather than application of density correction to low-density urine samples in recommended.