Mammographic examinations of 169 patients with 172 biopsy-proved carcinomas, and of 194 healthy subjects, were interpreted independently and retrospectively by three experienced mammographers, initially as single-view oblique examinations and 6 months later as two-view oblique-cephalocaudal examinations. For the single-view examinations of the cancer patients, 67% of the cancers were correctly recommended for biopsy, additional views were requested for 23%, and a "negative" interpretation was made for 10%. For the single-view examinations of healthy subjects, biopsy was recommended for 7% and additional views were recommended for 32%. For the two-view examinations of women with cancer, 80% of the cancers were correctly recommended for biopsy, additional views were requested for 4%, and a "negative" interpretation was made for 16%. For two-view examinations of healthy subjects, biopsy was recommended for 7% and additional views were requested for only 5%. The authors conclude that single-view screening should not be performed, because it would lead to an excessive number of "call-back" examinations of healthy patients, producing additional cost and anxiety that would outweigh any theoretical benefit.