OBJECTIVES: Health state utility values (HSUV) required for pharmacoeconomic evaluations are often unavailable from randomised controlled trials (RCTs). Instead some RCTs collect health related quality of life (HRQoL) data and mapping methods have been developed to derive utilities from such data for pharmaeconomic evaluations. Published mapping algorithms are either linear functions of detailed HRQoL item level data or nonlinear functions of domain scores, which require individual patient data as inputs. Researchers often do not have access to item scores or individual patient data, but are more likely to have summary domain score data available to them. The objective of this study was to propose methods to obtain HSUV from nonlinear mapping algorithms using summary HRQoL domain scores. METHODS: We used a linear Taylor approximation to a nonlinear mapping algorithmic function of summary HRQoL domain scores centred at the midpoint of a clinically important difference. We illustrate this for estimating utility values before and after disease progression, using summary domain scores of the FACT-G from an RCT of a targeted therapy for advanced cancer, as inputs to the response mapping algorithm estimated by Longworth. We compare our results with mapped utilities obtained from the best fitting (linear) item-level algorithm using individual patient data.  RESULTS: We found that the utility of stable disease was 0.783 with the linearised algorithm and 0.779 (95% CI: 0.763 - 0.796) with the best-fitting algorithm using IPD; utility post-progression was 0.747 and 0.725 (95% CI: 0.706 – 0.744), respectively. CONCLUSIONS: Linearisation of existing non-linear mapping algorithms may be used to obtain reliable HSUV for informing healthcare decisions, in the common situations where only summary HRQoL domain score data are available for research.