Content analysis was used to examine the consistency of responses in self-report data obtained by a semi-structured conversational interview. It was hypothesized that a majority of the seemingly inconsistent responses would prove to be consistent when seen from the respondents' points of view. The sample consisted of 215 female homicide offenders incarcerated in New York State correctional facilities or on active parole supervision in New York City. Results indicate that of the 24 respondents initially identified as having inconsistent responses, only one was regarded as inconsistent after the content analysis was performed and a coding scheme was developed to examine the apparent inconsistency from the respondent's point of view. The coding scheme developed also predicted homicide-related variables more effectively than did the original coding scheme developed for this sample. Suggestions for improving the construction of interview questions are also presented. (Contains 3 tables and 22 references.) (Author)