The cost-effectiveness of breast cancer screening that includes breast self-examination (BSE) has not been investigated, in part because the costs and effects of programs to teach BSE are not well known. I used data from a randomized, controlled trial and national surveys to compare the benefit, cost, and cost-effectiveness of two one-year, nurse-led programs to teach BSE in a medical practice setting. One program (Mammacare, or MAM) used manufactured silicone models to develop tactile skills; the other program (Traditional, or TRAD) stressed knowledge and examination techniques. I defined benefit as the marginal one-year increase in the number of women in an "average" adult primary care office practice who performed competent and frequent BSE. I defined cost as the marginal increase in the resources used to teach BSE and in the resources used as a result of performing BSE. Cost-effectiveness equaled the cost per competent, frequent BSE examiner added. The "average" practice was derived based on estimates from national surveys of physicians and medical practices. The MAM program's benefit was fivefold greater than the TRAD program's; its total cost was more than threefold greater; and its cost per competent, frequent examiner added was 32% lower. Yet, regardless of the program, one year following teaching only 15%-28% of women were competent, frequent performers. Cost-effectiveness was sensitive to the definition of benefit. Cost-effectiveness was also sensitive to some costs, but generally these sensitivities did not alter the relative cost-effectiveness of the two programs. The cost of teaching BSE was substantial. Benefit, cost, and cost-effectiveness varied by program.(ABSTRACT TRUNCATED AT 250 WORDS)