Abstract Background: Lower enrolment of minorities into research studies has been reported frequently. Most studies have little information about nonparticipants, making it difficult to identify characteristics associated with enrolment and how they might vary by race. Methods: Women who had previously participated in a population-based, case-control study of breast cancer in North Carolina were invited to enrol in a cancer genetics registry. Detailed questionnaire data on sociodemographic characteristics and cancer risk factors were available for all women. We compared characteristics of women who agreed to be in the registry with those who were deceased, were unlocatable, or declined enrolment. Unconditional logistic regression analyses were done to identify predictors of enrolment. Results: Enrolment rates were markedly lower among African Americans than Whites (15% and 36%, respectively) due to both lower contact rates (41% versus 63%) and lower enrolment rates among those contacted (37% versus 58%). Logistic regression models suggested that racial differences in enrolment were not due to socioeconomic characteristics or other cancer risk factors; race was the only significant predictor of enrolment in multivariable models (odds ratio 0.41, 95% confidence interval 0.23-0.72). Conclusions: Although all women had previously taken part in a research study, African American women were less likely to enrol in the cancer genetics registry than White women. A possible explanation of these findings is that studies of genetics may present particular concerns for African Americans. Further research is needed to identify attitudes and issues that present barriers to participation among minorities.