stories. Furthermore, there may have been additional concerns that recounting the violence would be more traumatizing for children and therefore children may have been less often selected for interview. Third, this study was limited by the open, self-reporting format and by a failure to ask follow-up questions. This limitation led to some variables frequently being coded as “Non-Specified”. As an example, 42% of all interviewees identified the perpetrator as “assailant” only and no further identifying information could be gathered from the record. Forty-two percent represents a high proportion of the sample and it would have been extremely useful to have collected more specific information. As a result, we were only able to study the patterns of attack according to types of perpetrators and speculate, based on cross tabulation of data, that the “Non-Specified” perpetrators as a group showed several important similarities to the armed combatants. The selfreporting nature of the survey also meant that outcomes such as physical, psychological and social problems could not be adequately quantified. Failure to volunteer a specific symptom or outcome does not imply its absence. While there were a number of women who self-reported symptoms of STIs, pregnancy resulting from rape, mental health symptoms and losses including spousal abandonment and death of family members, these numbers are most certainly underestimates. To have analyzed these data and reported them here would have served only to understate the physical, psychological and social consequences experienced by rape survivors. Finally, several translations were required before the analysis of these data. Most women in South Kivu Province speak Kiswahili or Mashi and the majority of them would have described their sexual violence to the in-take officer using one of these two languages. The in-take officer then translated the responses to French and recorded the data in French. To complete the analysis, all data were then translated into English by one of the investigators. As with all translation, these steps introduce the potential for error. However, the in-take officers were fluent in Kiswahili, Mashi and French and the study investigators were comfortable translating from French to English and were aided in their translations by the Panzi Hospital translator. Additionally, cultural differences have the potential to introduce error into the analysis. To limit this potential source of error, the results were discussed with local staff at Panzi Hospital who provided cultural background and context. Mealtime at Panzi Hospital (Susan Bartels, HHI) 48