The objective of this study was to develop an instrument to help clinicians inform patients with breast cancer of risks and benefits of adjuvant chemotherapy as derived from clinical trials and to help the informed patient decide whether she prefers treatment or no treatment. The instrument consists of a visual aid (called the decision board) and written material. It provides detailed information on a patient's choices (chemotherapy or no chemotherapy), outcomes (recurrence or not), probabilities of outcomes and their meaning, and quality of life associated with treatment choice and outcome. The validity and reliability of the instrument were evaluated in 30 healthy female volunteers. It was first administered using standard estimates of recurrence for node-negative breast cancer (15% risk of recurrence without treatment, which is reduced to 10% with chemotherapy). A preference for treatment (or no treatment) was then elicited. The validity was evaluated by changing the information provided on risks and benefits and determining whether the preference changed in a predictable manner. To test for reliability, the instrument was administered 2 weeks later. Seventeen women chose chemotherapy and 13 chose no chemotherapy. In the former group, 14 women (82%) switched preference when the magnitude of benefit was reduced, and 16 (94%) switched when the toxicity of treatment was increased. For those women who chose not to receive chemotherapy, 12 (92%) switched when the benefit was increased and 100% switched when toxicity was eliminated. The reliability was excellent (kappa = 0.86). The instrument has been used to elicit treatment preferences in 37 newly presenting patients with high-risk, node-negative breast cancer and has been found to be acceptable and helpful to the patient.