To compare self-report of connective tissue disease (CTD) with medical records, subjects were selected from 395,543 female health professionals with and without breast implants who reported CTD on mailed questionnaires from 1992 to 1995. The authors identified 220 women with breast implants (exposed) who self-reported CTD and a random sample of 879 women without breast implants (unexposed) who also self-reported CTD, matched by age and date of diagnosis. Medical records were reviewed using classification criteria from the American College of Rheumatology or other published criteria. After up to three requests and a telephone call, 27.7% of the women provided consent for medical record review. Exposed women appeared somewhat more likely (33.2% vs. 26.3%, p = 0.04) to provide consent. Using medical record reviews for 90% of the women who provided consent, confirmation rates of definite CTD were similar among the exposed and unexposed (22.7% vs. 24.0%, p = 0.83). This study demonstrates the difficulty of obtaining consent for medical record review of CTD reported to have occurred years ago in women with and without breast implants. Confirmation rates were low but were similar in exposed and unexposed. Despite the fact that the study had low participation rates, the data suggest that relative risk estimates for any definite CTD among women with breast implants compared with women without breast implants would be similar in analyses of self-reported or medical record-confirmed cases.