We conducted a community-based cluster sample survey of rubella sero-epidemiology in Addis Ababa, Ethiopia in 1994. Among 4666 individuals for whom complete data were available, rubella antibody prevalence was 91% (95% confidence interval: 90, 92). On multivariable analysis, seroprevalence was lower among individuals who were resident in Addis Ababa for 1 year or less. Approx. 50% seroprevalence was attained by age 4 years, and the estimated average age at infection was 5·2 years. The highest age-specific force of infection was estimated to occur in 5- to 9-year-olds. The early age at infection corresponded with a low estimated incidence of congenital rubella syndrome (CRS) of 0·3 per 1000 live births, equivalent to nine cases of CRS in 1994. The predicted critical level of immunity for elimination of rubella via vaccination was 85–91%, requiring 89–96% coverage with a vaccine of 95% effectiveness. Unless very high coverage of rubella vaccine could be guaranteed, the introduction of childhood vaccination could increase the incidence of CRS in Addis Ababa.