OBJECTIVES: To help facilitate economic evaluations of interventions for treating cancer, we estimated utility indices for the frequently used cancer-specific instrument of quality of life, by mapping it onto each of the HUI-2 and HUI-3 general health preference-based indices. METHODS: A sample of 367 cancer patients from the Vancouver Cancer Centre completed the FACT-G, HUI-2 and HUI-3 health-related quality of life questionnaires. Models of the relationships between the FACT-G and each of the preference based indices were estimated using regression analyses. We examined three alternative modeling approaches: ordinary-least-squares (OLS); generalized-linear-modeling (GLM) using a Gaussian distribution and log link; and censored-least-absolute deviations (CLAD). The performance of the models was assessed in terms of how well the responses to the cancer-specific instrument predicted utilities from each of the preference-based instruments using the mean absolute error (MAE) and root mean square error (RMSE). We adjusted for the socio-demographic and clinical characteristics of patients. RESULTS: Results from the final models of the three approaches did not differ significantly. Physical, functional and emotional subscales of the FACT-G were significant predictors of the HUI-2 and HUI-3 utility scores. The root mean square error for the HUI-3 was lower, suggesting better predictions than for the HUI-2.  The OLS/GLM predicted HUI-2 and HUI-3 scores matched the observed values more closely than the CLAD. CONCLUSIONS: There is potential to estimate both HUI-2/HUI-3 utilities using responses from the FACT-G cancer-specific measure of quality-of-life, even though it was not designed as a utility instrument. Our results suggest that it is possible to estimate Quality-Adjusted-Life-Years from studies where only cancer-specific instruments have been administered.