OBJECTIVE: The objective was to combine, in one, paper-based, self-completion instrument, the strengths of two approaches to assessing health-related quality of life: the sensitivity and relevance of the disease-specific approach and the generalizability and decision support power of the utility approach. The instrument was developed for erectile dysfunction (ED). METHODS: The instrument has two linked stages, each with a visual analog scale (VAS). VAS1 measures preferences for the patient's self state and for disease marker states. VAS2 measures preferences on the traditional dead-healthy 0-1 scale. The instrument was used in a double-blind, placebo-controlled, 12-week clinical trial (n=169). At subject recruitment, it was administered twice, one week apart. All results reported here are based on that data, except responsiveness, which is based on the 12-week data for both treatment groups combined. RESULTS: Feasibility: Completion rates were approximately 90% overall and approximately 95% for key data. Error rates were 2%. Most patients required minimal or no assistance. Reliability: One-week, test-retest reliabilities of all preference scores, as measured by intra-class correlation, were strong (>0.50). The key score, self-state on VAS2, had the highest reliability (0.70). Face validity: Scores were ranked in the expected order. Discriminative validity: Self-state scores were positively and statistically significantly correlated with disease severity. Pearson correlation coefficients were 0.37 on VAS1, 0.24 on VAS2. Responsiveness: Change over 12 weeks in self-state scores were positively and statistically significantly correlated with change in disease severity. Spearman correlation coefficients were 0.44 on VAS1, 0.37 on VAS2. CONCLUSIONS: We have developed a preference-based, disease-specific instrument for ED. The instrument provides preference scores for the patient's self state and for clinical marker states on a disease-specific scale and on the traditional 0-1 dead-healthy generic scale. The performance of the instrument in this initial application was encouraging. The instrument demonstrated feasibility, reliability, validity and responsiveness.