This study analysed sex ratios at birth (defined as the number of male births per 100 female births) using data on children ever-born from three censuses conducted in Ethiopia in 1984, 1994 and 2007. The results showed very high values by any standard, with an average of 108.4 for a sample of some 8.2 million births, with somewhat lower values in urban areas. Analysis of socioeconomic correlates revealed that the sex ratio varied very much by household wealth, from about 110 for very poor women to about 102 for wealthier women. The high value of the sex ratio at birth in Ethiopia could be explained by poverty, used as a proxy for poor nutritional status. In multivariate analysis, the effects of living in urban areas and of maternal education were less important than household wealth. Among the many ethno-linguistic groups, the Nilotic family had higher sex ratios than other groups. The results were confirmed using data from DHS surveys conducted in the country, and by the analysis of children still living at time of census.