The study answers questions dealing with extent of response error, random versus systematic response errors, the reasons for response errors (lapses in memory and motivational factors that lead to underreporting or overreporting), and the extent to which the survey response is an adequate substitute for actual data. The sample for the investigation was drawn from the ex-addict group in the Vera Institute of Justice Supported Employment Experiment. From the total research sample, a subsample of 79 who had a complete arrest record on file with the New York Police Department was selected for the study. As a part of regular quarterly interviews, respondents were asked about their arrest histories. The self-report measure was designed to be comparable to the criterion measure for police arrest records. The response error in self-reports of arrests was constructed by subtracting the actual number for each person. The independent variables selected as predictors of response error were of three kinds: (1) the potential of the question to motivate response errors (as in the generally perceived seriousness or undesirability of crimes), (2) question difficulty (or recall difficulty inherent in the question), and (3) personal motivation (personal beliefs about undesirability or desirability of an arrest record). Only 10 of 79 subjects gave accurate responses. Almost half gave responses that were within plus or minus two arrests of the true number. Twenty percent made errors of plus or minus five or more arrests. Implications of the results for the conduct of surveys are discussed. Tabular and graphic data, 2 notes, and 25 references are provided.