This paper studies the relationship between civil war and HIV/AIDS in Burundi. It contributes to the empirical literature by providing micro level evidence using an identification strategy based on original data on the dynamics of rebel movements. The presence of exit and entry points from and to rebel safe havens is used to generate exogenous variation in conflict intensity. These points are plausibly assumed to serve as starting or end points for rebel attack, but are not directly related to HIV/AIDS or correlated with unobservables. The case of Burundi provides fruitful grounds of analysis, as seroprevalence rates are heterogeneous across the country, the serological and conflict data for Burundi is of good quality and conclusions are likely to serve as valuable insights in Burundi and other fragile countries with similar HIV/AIDS policy agendas. OLS, instrumental variable and binary response model results indicate that within provinces in Burundi there is no clear-cut relationship between local conflict intensity and seroprevalence, condom knowledge and use, knowledge of test opportunities and actual test taking, or rape. Findings suggest that although HIV/AIDS is a general development priority, it is not as urgent a post-conflict priority as commonly assumed.