Studies of civil wars often allude to the potential for problems of memory to mar observational data collected through surveys. The validity of survey respondents’recollections is a particular concern for the field of civil war studies given the trend towards micro-level research. Researchers making use of in-depth interviews might be ableto assess the accuracy of respondents’ memories on a case by case basis. But no such tool seems readily available when collecting systematic data through survey research. As Arjona (2010: 398) notes, when it comes to conflict, the dilemma is that even when researchers fully understand the distorting effects of memory, there may simply be no source of information on events and behavior of interest other than survey or interview respondents. How, then, can researchers defend the validity of inferences based on retrospective survey data on conflict? To what extent do such data create insurmountable challenges to inference?