The survey was conducted in April-May 1985, using a version of random digit dialing that eliminates nonworking banks of telephone numbers. Female interviewers completed interviews, averaging 20 minutes, with 315 women aged 18 to 65 who lived in metropolitan Toronto, Canada. Respondents were randomly assigned to either the closed or open condition questionnaire that asked about physical abuse by a husband, boyfriend, or date. Of the group, 155 completed the closed questionnaire and 160 the open version. The study found no significant differences in the ability of closed versus open questions to elicit disclosures of abuse, although the open format tended to produce more detailed reports. An unanticipated finding was that 21 percent of the self-reported victims revealed their victimizations, not when initially asked, but later in the interview when asked 'Have you had any experiences as a victim of abuse that I have not asked about?' This finding suggests that respondents should be given more than one chance to reveal such information in surveys on abuse. (Author abstract modified)