OBJECTIVE: Determine feasibility of shared decision-making programmes in fee-for-service hospital systems including physicians' offices and in-patient facilities. DESIGN: Survey and participant observation. Data obtained during Phase 1 of a patient outcome study. SETTINGS AND PARTICIPANTS: Three hospitals in Michigan: one 299-bed rural regional hospital, one 650-bed urban community hospital, one 459-bed urban and suburban teaching hospital. All nurses and physicians who agreed to use the programmes participated in the evaluation (n = 34). INTERVENTION: Two shared decision-making(R) (SDP) multimedia programmes: surgical treatment choice for breast cancer and ischaemic heart disease treatment choice. MAIN OUTCOME MEASURES: (1) clinicians' evaluations of programme quality; (2) challenges in hospital settings; and (3) patient referral rates. RESULTS: SDP programmes were judged to be clear, accurate and about the right length and amount of information. Programmes were judged to be informative and appropriate for patients to see before making a decision. Clinicians were neutral about patients' desire to participate in treatment decision-making. Referral volume to SDPs was lower than expected: 24 patients in 7 months across three hospitals. Implementation challenges centred on time pressures in patient care. CONCLUSIONS: Productivity and time pressure in US health care severely constrain shared decision-making programme implementation. Physician referral may not be a reliable mechanism for patient access. Possible innovations include: (1) incorporation into the informed consent process; (2) provider or payer negotiated requirement in the routine hospital procedure to use the SDP as a quality indicator; and (3) payer reimbursement to professional providers who make SDP programmes available to patients.