Debate exists about the influence of screening on breast cancer surgery. We compared rates and odds of mastectomy and breast-conserving surgery (BCS) during 1994-1999 between Northern Ireland, which has had organized mammographic screening (age group 50-64) since 1993, and the Republic of Ireland, where large-scale screening did not begin until 2000. Trends in BCS were similar in both populations: significant increases in BCS rates for age groups 20-49 and, especially, 50-64, and significant increases in mastectomy rates for age group 65+ only. Use of BCS among surgical patients was significantly higher in Northern Ireland (46%) than the Republic of Ireland (35%) and, in Northern Ireland, was significantly higher in age group 50-64 than other age groups. Geographic and age-related variations were substantially, but not wholly, explained by tumour characteristics and screen-detection status. Among Northern Ireland women aged 50-64, BCS use among screen-detected cases, even after adjustment for tumour characteristics, was 40-50% higher and increased more rapidly during 1994-1999 than among other cases. However, trends in BCS in Northern Ireland were not explained by changes in screen-detection status or tumour characteristics. Our findings are consistent with expectations that mammographic screening, together with modern treatment guidelines, should lead to reduced use of mastectomy and increased (proportional) use of BCS. But similar trends occurred in the absence of large-scale screening (in the Republic of Ireland), and surgical treatment of screen-detected cases did not reflect tumour characteristics alone.