OBJECTIVE:To develop a rating scale to elicit and quantify patient preferences for different dosing frequencies with bisphosphonate therapy in osteoporosis. METHODS: Expert interviews generated information for developing descriptions of an osteoporosis health state along with four different treatment scenarios for bisphosphonates: daily, weekly, monthly, and monthly plus a support programme. A rating scale questionnaire was used to elicit relative values for the health state and each scenario on a 0-100 scale.Face and content validity, acceptability and comprehension were tested with experts (n=3) and osteoporosis patients (n=22). Telephone administration of the rating scale was assessed in a pilot study of 50 bisphosphonate users. Construct validity was evaluated in 491 women with or at risk of osteoporosis who were: naïve to (n=212), currently taking (n=225) or lapsed (n=54) bisphosphonate users. RESULTS:The questionnaire demonstrated good face and content validity, was acceptable to experts and patients, and suitable for telephone administration taking 5-10 minutes to complete. Construct validity was demonstrated by: (i) Lower valuation for health state compared with treatment scenarios (median values: health state=20, daily=40, weekly=70, monthly=80, and monthly with support programme=85); (ii) Non-osteoporotic women gave lower valuations for health state than women with osteoporosis (median value=10, one-way ANOVA, F=4.80, p = 0.009); (iii) Lapsed weekly users gave lower valuations to weekly scenario (median value=50) than naïve or current users, one-way ANOVA, F=17.68, p<0.0001; (iv) 94% of lapsed and 85% of current users who felt they had to take bisphosphonates too often, preferred monthly treatment; (v) monthly dosing with support programme was the most highly valued scenario by all patient groups (median value=85). CONCLUSIONS:The preference questionnaire is a valid and acceptable method for quantifying patient preference for dosing frequency with bisphosphonate therapy. Overall, bisphosphonate treatment scenarios with lower dosing frequencies were valued more highly.