In 1987 a pilot project that involved joint police and social services investigation of suspected cases of child sexual abuse was begun in an English county, and the authors of this article evaluated its first year of operation. During the 12-month evaluation, 83 referrals were received. Forty-five investigations included at least one video- recorded interview; 11 involved 2 with the same children, and 11 involved a videotaped interview with a sibling, totaling 67 video-recorded interviews. Allowing for technical failure and refusals of consent, 44 recordings were available for analysis. A pro forma for quantitative analysis of interview procedure was devised after close, repeated inspection of a subset of recordings. It was applied to all the available recordings. Variables in the pro forma included the interview's length, who was present and any temporary absences during the interview, whether and how anatomically detailed dolls were used, which investigator began the interview and whether either took a leading role, the verbal responsiveness and demeanor of the child, whether the child was allowed time to speak and ask questions, what sort of information was given the child about the referral and what would happen next, and whether assurances about confidentiality and responsibility were given. Other information pertained to the extent to which the interview was structured, the questioning techniques used, procedures for establishing the child's terminology for body parts and significant others, and the nature of physical contact between investigators and the child. The authors note that investigator training currently emphasizes that the best means of obtaining reliable evidence is the careful conduct of the interview. The authors argue that similar emphasis must be given the analysis of the dialog as a product of the context of an interaction between investigators and victims. They maintain that techniques for the interpretation of research interview data can assist practitioners in assessing the quality of investigative interview evidence. The study illustrates how this may be done with several brief extracts from investigative interviews with very young victims. 1 table and 41 references