Sexual violence is believed to be widespread during war. Yet empirical evidence concerning its prevalence is often limited. Victims, out of feelings of shame or fear, underreport this form of violence. We tackle this problem by administering a list experiment in a representative survey in Sri Lanka, which is only recently recovering from an ethnic civil war between Sinhalese and Tamils. This unobtrusive method reveals that around 13 percent of the Sri Lankan population has personally experienced sexual assault during the war—a prevalence ten times higher than elicited by direct questioning. We also identify vulnerable groups: Tamils who have collaborated with rebel groups and the male-displaced population suspected of collaboration with the Liberation Tigers of Tamil Eelam. Our experimental evidence thus lends support to reports on the asymmetric use of sexual violence by government forces, qualifies conventional wisdom on sexual violence during war, and has important implications for policy.