The aim of the present study was to investigate the quality of the interview process that was used to obtain data on the prevalence of elder mistreatment in order to improve the quality of this process in future research studies. Audiotapes of interviews with 143 victims of elder mistreatment were transcribed into verbatim protocols that were subsequently coded and analyzed. The results of the present study show that despite the fact that all interviewers were carefully selected, intensively trained, and supervised during the interview period, 4.2% of the questions were skipped, and in 4.4% of the interactions no usable answer was obtained. However, in all of these cases the interviewer did register an answer, the validity of which is questionable. Considerable differences were found between interviewers with respect to their performance and also between the various questions. Finally, suggestions are made to improve interviewer behavior.