This study aimed to identify factors that predicted attendance at a relocatable screening mammography service in a rural centre in Victoria. A cohort design was used whereby 180 women from the target population were interviewed by telephone two weeks before the service moved to the area for a 10-week period of operation. Attendance data were ascertained from service records. Fifty per cent of the sample attended the service. Significant predictors of attendance were: mammographic history, with women who reported previous screening mammography being less likely to attend than women who had not had a previous mammogram (odds ratio (OR) 0.38, 95% confidence interval (CI) 0.17 to 0.83); perception of personal risk for breast cancer, with women who perceived at least some risk being more likely to attend than women who perceived no risk (OR 2.73, CI 1.07 to 6.99); stated intention of attending (OR 2.01, CI 1.49 to 2.71); knowing the correct location of the service (OR 3.08, CI 1.37 to 6.89); and education, with higher education being associated with a lower likelihood of attending (OR 0.65, CI 0.44 to 0.96). Our study raised some issues, including the high prevalence of rural women who reported a previous screening mammogram, although BreastScreen services had not previously been available in their area; factors underlying perceptions of personal risk for breast cancer; and the generalisability of our finding of an inverse relationship between higher education and attendance for screening.