A positive family history increases the risk for breast cancer which oft en occurs at a much younger age than in the general population. We stud ied whether surveillance of these women resulted in the detection of bre ast cancer in an earlier stage than in symptomatic patients with a famil y history. Between January 1994 and April 1998, 294 women with 15-25% r isk (moderate), mean age:43.3 (22-75) years, were screened with a yearly physical examination and mammography from 5 years before the youngest ag e of onset in the family and 384 women with >25% risk (high) for breast cancer, mean age: 42.9 (20-74) years were screened with a physical examination every 6 months and yearly mammography. From September 1995 breast magnetic resonance imaging (MRI) was also carried out for 109 high risk women where mammography showed over 50% density. 26 breast cancers detected under surveillance were significantly more often found in an early T1N0 stage than the 24 breast cancers in patients with a family history referred in that period because of symptoms: 81 versus 46% (P=0.018). Patients under surveillance were also less frequently node-positive than the symptomatic group: 19 versus 42% (P=0.12). 20 patients with a family history referred by our national screening programme in that period had 21 breast cancers detected, 81% in stage T1N0 and 5% node-positive, which was comparable to the results in our national screening programme T1N0 66%, N+ 24% resulting in a 30% reduction in mortality. The incidence in women under surveillance was 10.1 per 1000 in the 'high' risk group and 13.3 per 1000 in the 'moderate' risk group. Expected incidence in an average risk population aged 40-50 years is 1.5, expected if the group consisted of only gene carriers 15 per 1000. 23% of the breast cancers in the surveillance group were detected at physical examination, but occult at mammography. 38% were detected at mammography and clinically occult. Breast MRI (in the subgroup) detected 3 occult breast cancers. The results of this study show that women with a family history benefit from surveillance as breast cancer was detected significantly more often in a favourable T1N0 stage and a mortality reduction comparable to that obtained in our national screening programme may be expected also in women <50 years of age. Both physical examination and mammography contribute to this result, but the former in this study only contributed in women before menopause. Starting surveillance some years before the youngest age of onset in the family may result in higher detection rates. Screening with MRI can detect breast cancers, still occult at physical examination and mammography.