The report reviews some features of crime which affect our ability to measure it accurately, including the relative infrequency of serious victimization, the skewed distribution of victimization in the population, and the furtive character of crime. The third chapter addresses issues related to how survey questionnaires are designed to define and measure victimization. It examines the events orientation of victimization surveys, the assumption that crimes always are discrete incidents rather than continuous social processes, and the utility of measures of criminal activity abstracted from their social context. The fourth chapter reviews specific measurement problems: limited distribution of knowledge of incidents, forgetting or inaccurate recall of events, and differential productivity of survey respondents. The next chapter reviews three procedural issues which affect estimates of victimization rates: problems of panel bias and attrition, differences between telephone and inperson interviews, and interviewer effects. The final chapter summarizes the current state of the art in this area and discusses possible future developments in victimization survey methodology. (Author abstract )