In a two stage retrospective survey on child sexual abuse (CSA), a majority of CSA incidents were reported at both postal and interview stages of the survey. Incidents that involved genital contact were most likely to be reported at both stages. A significant number of women reported abuse by a close family member in the postal questionnaire but not at interview. In contrast, incidents mentioned only at interview were more likely to be carried out by a stranger. At interview, many women who replied negatively to a general screening question on CSA went on to report abuse in response to detailed descriptive questions. This was especially true for noncontact experiences. Increasingly restrictive definitions of CSA lowered the prevalence rate for CSA under the age of 16 years from 34.4% to 19.7%. The implication of these findings for the design of future studies is discussed.