OBJECTIVES: The St George’s Respiratory Questionnaire (SGRQ) is a health related quality of life (HR-QoL) measure for respiratory conditions including asthma; however it does not have associated utility values required to generate quality adjusted life years (QALYs) used in economic evaluation. A mapping algorithm has previously been estimated to generate EQ-5D-3L values from the SGRQ in a COPD population. To improve precision in mapping utility values for asthma patients, we developed a mapping function between the SGRQ and EQ-5D-5L specifically for the severe asthma population. METHODS: Data from the cross-sectional Identification and Description of Severe Asthma Patients (IDEAL) Study (n=748) were used in this post-hoc analysis to map between the two measures via OLS, Tobit and two-part models (TPM). Model specifications including SGRQ item, total, or dimension scores were compared. The models were tested on a validation sample from the same cohort; comparisons of observed and predicted scores and the overall mean absolute error (MAE) were assessed to determine performance.  RESULTS: OLS and Tobit SGRQ item models were better at predicting the mean EQ-5D-5L values than SGRQ total or dimension score models or the TPM. Tobit SGRQ item models had the lowest MAEs (0.111 – 0.115) which are lower than those for the existing EQ-5D-3L algorithm (0.124). All mapping models showed poorer fit and over-prediction at the lower end of the EQ-5D distribution. This was a consequence of limited observations at this end of the scale. CONCLUSIONS: Tobit models based on SGRQ item scores had the lowest overall MAEs; the results suggested an improved level of precision as compared to the existing algorithm. However, none of the models tested performed well at the lower end of the EQ-5D distribution. Funded by GSK (HO-15-16182)