The percent population attributable risk (AR) for breast cancer was estimated in relation to education, family history of the disease and some reproductive and hormonal factors, using data from a case-control study conducted between June 1991 and February 1994 in 6 Italian centres on 2,569 histologically confirmed incident breast cancer cases and 2,588 controls, admitted to hospital for a wide range of acute, nonneoplastic, non-hormone-related diseases. On the basis of multivariate odds ratios, a high level of education accounted for 20% of cases, elevated age at first birth and nulliparity for 38% and a family history of breast cancer in first-degree relatives for 7%. Education and nulliparity and age at first birth together explained 47% of all breast cancer cases, and the combination of these 2 factors plus a family history of the disease explained 50% of cases. In pre-menopausal women a high level of education accounted for 31% of all breast cancer cases, older age at first birth for 44% and the combination of the 2 factors for 49%. In post-menopausal women the corresponding values were 13%, 31% and 42%; further addition of risk associated with family history of the disease explained 52% of pre-menopausal cases. In post-menopausal women older age at menopause and the use of hormone replacement therapy accounted for 15% and 2% of breast cancer cases, respectively. The combination of risks associated with a high level of education, old age at first birth and nulliparity and older age at menopause accounted for 51% of cases; further inclusion of risk associated with use of hormone replacement therapy explained 52%, and the AR resulting from these 4 risk factors combined plus a family history of breast cancer was 56%. Thus, a few selected and well-identified risk factors explain about one-half the breast cancer cases in this Italian population. Int. J. Cancer, 70:159-163, 1997.r 1997 Wiley-Liss, Inc.Breast cancer is the most common neoplasm in women in developed countries, though its rates vary considerably across different populations (Harris et al., 1992;Ewertz and Duffy, 1994;Levi et al., 1994;Lipworth, 1995). A role of several hormonal factors in terms of relative risks has been consistently reported (Henderson, 1993;Kelsey et al., 1993;Lipworth, 1995), and although the strength of the associations for each risk factor or their combinations is not very elevated, their variation may well at least partly explain the national differences in breast cancer incidence and mortality (La Vecchia and Pampallona, 1986;Harris et al., 1992).Since attributable risks (AR) are a function of both odds ratios (OR) and prevalence of risk factors in the population, they tend to vary in different countries (Harris et al., 1992). A prospective study of the American Cancer Society (ACS) (Seidman et al., 1982) found that 21% of cancers diagnosed at ages 30-54 years and 29% of those diagnosed at ages 55-84 years could be attributed to one or more selected risk factors simply divided as ''low'' and ''high'' risk. An American case-c...