SummaryOBJECTIVE To assess the utility of oral fluid compared with serum for the determination of age-prevalence of rubella-specific antibodies in an urban African community setting. METHOD Paired serum and oral fluid samples were collected from 439 individuals aged 0-49 years in Addis Ababa, Ethiopia, as part of a larger seroepidemiological survey in 1994. Oral fluid was sampled using a simple sponge device that was well accepted by subjects of all ages; venous blood was collected by Vacutainer system. We measured rubella-specific antibodies in serum by the Radial Haemolysis (RH) test, supported by two confirmatory assays, and in oral fluid by IgG antibody-capture radioimmunoassay (GACRIA). RESULTS Sensitivity and specificity of oral fluid results compared to serum were 89% and 76%, respectively. Sensitivity declined from 96% in age group 0-19 years to 90% in age group 20-29 and 78% in age group 30-49. Specificity was 86% in 0-9 year olds contrasting with 61% in older groups (10-49 years). The positive predictive value of an oral fluid sample was high in all age groups (range 92-100%), while the negative predictive value declined from Ն80% in those aged Ͻ10 years to Ͻ10% in those aged Ն30 years. Serum confirmatory tests suggested a proportion of false serum RH negatives, increasing with age, indicating a need to standardize serum as well as oral fluid tests. CONCLUSION In the community setting of a developing country, oral fluid surveys could be useful to estimate age-prevalence of rubella immunity and identify rubella-susceptible children for follow-up. Further work is required to simplify assays and sample processing, improve assay sensitivity and estimate assay specificity more precisely, and compare and standardise collection methods suitable for surveillance of a variety of childhood viral infections.keywords oral fluid, rubella-specific antibodies, age-prevalence, Ethiopia correspondence Dr DJ Nokes,