BACKGROUND In Japan rubella vaccination is generally done once during a lifetime, and the vaccination rate decreased after a revised vaccination law in 1995. History of rubella or vaccination may still be unreliable. Testing for rubella antibody is significant to prevent the occurrence of congenital rubella syndrome. However, the collection of blood samples to detect antibodies from young children is invasive and difficult. METHODS For this study we obtained 853 matched serum and urine samples from 904 healthy students 10 or 14 years of age in the Ibara and Yoshii districts of Okayama, Japan, for a comparison of antibodies for rubella in the matched samples. The serum and urine antibodies were measured with hemagglutination-inhibition and enzyme-linked immunosorbent assays, respectively, and with our urine-based antibody test. RESULTS The sensitivity, specificity and concordance rates of this urine-based antibody test were 96, 99 and 97% based on the serum antibody results of both assays. The coefficiency was 0.627 between the titers of the urinary and serum antibodies by the enzyme-linked immunosorbent assay. The urinary antibodies were stable for at least 5 months at 4 degrees C and 25 degrees C. CONCLUSIONS Urine-based assay methods are helpful not only because they avoid the invasive approach of venipuncture but also because unprocessed urine specimens can be used and urinary antibody is stable for a long period. Therefore this test is suitable for screening. In addition protective amounts of rubella antibody in blood can be reliably assessed by means of urine samples.