OBJECTIVES: Mapping from disease-specific to utility measures is a research area gaining increasing attention. However, few studies have assessed the longitudinal validity of mapping models and none, to our knowledge, have compared models derived from baseline and post-intervention patient data. This study examined models derived from time-differing patient data, for mapping the Modified Health Assessment Questionnaire (MHAQ) on to the EQ-5D. METHODS: A total of 120 rheumatoid arthritis patients (60.0% female, mean age 59.0) completed the MHAQ and EQ-5D at baseline, and 3, 6 and 12 months after administration of an anti-TNFa or another biological agent. OLS regression produced mapping equations from baseline and post-intervention data. Model predictive ability and explanatory power were assessed by root mean square error (RMSE) and adjusted R2 respectively. Pearson’s r and intraclass correlation coefficient (ICC) assessed association and level of agreement between predicted and reported utilities. RESULTS: R2 (baseline, 3, 6 and to 12 months) was 0.452, 0.418, 0.541 and 0.413 respectively, whereas prediction errors were 13.0%, 11.1%, 8.1% and 7.2%. All equations produced a range of scores comparable to those achieved by the standard EQ-5D scoring algorithm, as well as strong correlations and agreement between reported and predicted utilities. The baseline and 12-month models appeared to under- and overestimate EQ-5D scores respectively, whereas the 6-month model generated the smallest differences, typically less than the minimally important difference for the EQ-5D (0.03). CONCLUSIONS: Baseline data, which typically corresponds to lower patient HRQOL, may give mapping equations with compromised predictive ability, compared to models generated from post-intervention data, and this study showed that this factor is worth examining when longitudinal data are available. The next step in this line of research is to test QALY estimates and cost-utility increments derived from baseline and post-intervention mapping algorithms.