Purpose.This study examined the interviewing ability of benefit fraud investigators, specifically examining the effects of training in the PEACE model. This model, preferred by the British police, is a mnemonic for the interview process: planning and preparation, engage and explanation, account, closure, and evaluation.Methods.An analysis was undertaken of 99 audiotapes of real‐life interviews with benefit fraud suspects, comparing performance levels between trained and untrained investigators, using an assessment scale consisting of 56 behavioural and procedural elements adapted from Clarke and Milne (2001).Results.Some improvements were found in performance amongst the trained investigators notably with good practices being seen in the effective use of ‘open’ questions which facilitated increased information from suspects, in addition to a comprehensive fulfilment of the necessary legal requirements. Additionally, only rare occurrences were found of any unethical practices. However, the study also found little evidence of planning which was thought to affect interview content particularly in relation to their structure and flexibility. Concerns were also found with (i) shortfalls in rapport building, (ii) insufficient coverage of the points that required proving, (iii) failures to completely explore the suspect's motive and (iv) the lack of summarizing during the interview.Conclusions.The research found some performance improvement after interview training but this was insufficient to suggest a general trend towards increased professionalism. A need for further training was identified, both in the PEACE model and in other appropriate interviewing s