A cross-sectional sero-epidemiological study of bovine brucellosis was conducted between September 2005 and March 2006 in three separate agro-ecological areas of central Oromiya, Ethiopia. In this study, a total of 176 clusters (farms) and 1,238 animals were selected, using the one-stage cluster sampling method. Fifty-nine clusters and 423 animals were selected from the lowland areas; 58 clusters and 385 animals from the midland areas and 59 clusters and 430 animals from the highlands. Serum samples were collected from a total of 1,238 animals older than six months. The rose bengal plate test and complement fixation test were used as screening and confirmatory tests, respectively, to detect Brucella seropositivity. Questionnaires were also administered to 176 households to gather information on the farm and livestock. Results showed that the overall seroprevalence of bovine brucellosis at the individual animal level was 2.9% (low). The seroprevalence was 4.2% in the lowlands, 1.0% in the midlands and 3.4% in the highlands. The overall seroprevalence at the herd level was 13.6% (moderate). At the herd level, seroprevalence in the lowlands was 17%; in the midlands: 5.1%; and in highland areas: 18.6%. Logistic regression analysis revealed that the breed of cattle and the method of disposing of aborted foetuses and foetal membranes had a statistically significant effect on individual animal seroprevalence (p < 0.05). In lowland areas, the breed (p < 0.05), animal management system (p < 0.05), mating method (p < 0.05), herd size (p < 0.05) and source of replacement stock (p < 0.05) all had significant effects on individual animal seroprevalence.