OBJECTIVES: Rabies is a global problem, although it is often under-reported in developing countries. We aimed at describing the profile of patients presenting to health centres with animal bite injuries in Uganda, and use a predictive model to estimate the mortality of rabies at a national level. METHODS: We conducted a passive surveillance study in Uganda based in a random sample of health centres supplied with rabies vaccine to determine the characteristics of bite injury patients and establish the age and sex profiles of patients, the site of bites and their severity, wound management techniques and details of the vaccination course given. We also applied a decision tree model to the data to estimate the rabies mortality from the bite injury data using an established protocol. RESULTS: We found that most patients are bitten by dogs, and that a considerable proportion of these are young children, who are at greater risk of developing rabies in the absence of treatment due to the location of the bites they receive. From conservative parameter estimates, we estimate that in the absence of post-exposure prophylaxis (PET), 592 (95% CI 345-920) deaths would occur, and that if one dose of PET is sufficient for protection following a rabid animal bite, 20 (95% CI 5-50) deaths would occur annually. If a complete course of PET is required for protection following a rabid animal bite, up to 210 (95% CI 115-359) deaths would occur, as 41% of patients did not complete their course of PET. CONCLUSIONS: Active animal bite surveillance studies are required to improve our mortality estimates and determine the true burden of rabies in the Ugandan population. We emphasize the need for small-scale active case detection studies and improved data on the recognition of rabies in dogs as inputs for improving national-level estimates of rabies mortality.