On a population-based sample of 13,500 European breast cancer patients mostly diagnosed in 1996-1998 and archived by 26 cancer registries, we used logistic regression to estimate odds of conservative surgery plus radiotherapy (BCS+RT) versus other surgery, in T1N0M0 cases by country, adjusted for age and tumour size. We also examined: BCS+RT in relation to total national expenditure on health (TNEH); chemotherapy use in N+ patients; tamoxifen use in oestrogen-positive patients; and whether 10 nodes were examined in lymphadenectomies. Stage, diagnostic examinations and treatments were obtained from clinical records. T1N0M0 cases were 33.0% of the total. 55.0% of T1N0M0 received BCS+RT, range 9.0% (Estonia) to 78.0% (France). Compared to France, odds of BCS+RT were lower in all other countries, even after adjusting for covariates. Women of 70-99 years had 67% lower odds of BCS+RT than women of 15-39 years. BCS+RT was 20% in low TNEH, 58% in medium TNEH, and 64% in high TNEH countries. Chemotherapy was given to 63.0% of N+ and 90.7% of premenopausal N+ (15-49 years), with marked variation by country, mainly in post-menopause (50-99 years). Hormonal therapy was given to 55.5% of oestrogen-positive cases, 44.6% at 15-49 years and 58.8% at 50-99 years; with marked variation across countries especially in premenopause. The variation in breast cancer care across Europe prior to the development of European guidelines was striking; older women received BCS+RT much less than younger women; and adherence to 'standard care' varied even among countries with medium/high TNEH, suggesting sub-optimal resource allocation.