OBJECTIVE: Preference scores are needed for cost-effectiveness analyses that are particularly important for assessing expensive technologies such as intensive care therapies for ARDS patients. Preference data for ARDS patients can be generated using secondary analysis of descriptive HRQOL data, but it is not known which preference algorithm is to be preferred. Our objective was to compare two methods for deriving preference scores of ARDS survivors from SF-36 scores, one Quality of Well-Being based - Fryback et al.(1997) and the other Health Utilities Index based- Nichol et al.(2002) with directly rated Visual Analogue Scale (VAS) preference. Agreement between these approaches and their validity relative to other HRQOL measures were examined. METHODS: Data were collected from 43 ARDS survivors identified from three major hospitals in Twin Cities, Minnesota, from 1993 to 2001. The questionnaire included the VAS, the SF-36, Center for Epidemiologic Studies-Depression, life satisfaction and happiness, and Karnofsky Performance Index. Repeated measures ANOVA and post-hoc t-tests were used to analyze differences among preference scores. The intra-class correlation coefficient (ICC) was employed to assess agreement between the two scores. Spearman rank correlation coefficients (r) were used to measure the correlation between these preferences and other health outcomes. RESULTS: The mean (SD) preference scores were 0.603 (0.24), 0.632 (0.09), and 0.679 (0.18) for VAS, Fryback and Nichol methods, respectively. The agreement for VAS & Nichol was higher than that for Fryback & Nichol with ICCs of 0.7744 and 0.7247, respectively. Agreement for VAS & Fryback was poor with ICC of 0.4895. The Fryback score showed poor correlation with SF-36 Mental Component Summary score, life satisfaction and happiness (r <0.40); the Nichol and VAS preferences had higher correlations with these measures. CONCLUSIONS: Choice of preference algorithm impacts both the mean preference and its validity. In ARDS, the Nichol approach may be preferred.