OBJECTIVES: There is a growing concern that responses to questions on subjective scales will be inaccurate if certain groups of people systematically differ in their interpretation and use of the response categories - known as differential item functioning (DIF). It has been shown that it is possible to correct for DIF by using vignette responses collected externally to the main dataset of interest. We apply this approach to the EQ-5D-5L to demonstrate how this adjustment-method can be used in practice to obtain QALY measures that are comparable across different population groups. METHODS:  We adjust for DIF in the Multi Instrument Comparison (MIC) study (our main dataset of interest) using vignettes collected in an online survey of Australian respondents (the vignettes sample). We restrict our analysis in both samples to individuals aged 55 years and above (656 respondents in the MIC sample, and 914 in the vignettes sample). To adjust for DIF we use a special case of the HOPIT, where the likelihood functions index two different samples - the vignettes sample and the MIC data - which are linked through common parameters in threshold equations. DIF-adjusted profiles are obtained, and tariffs are applied to calculate DIF-adjusted EQ-5D indices. RESULTS: Differences in indices between the lowest and highest educated individuals increased from 0.054 before adjustment to 0.079 post DIF-adjustment, which is above a suggested minimally important difference (MID) of 0.074. The difference between employed individuals and those not employed increased from 0.093 to 0.141 after adjusting for DIF. Differences between married and non-married individuals also increased from 0.065 to 0.096, which is also above the MID. Differences across subgroups in the unadjusted and DIF-adjusted indices were not substantive across subgroups according to gender, migrant status or age group. CONCLUSIONS: Ignoring DIF could potentially bias conclusions regarding sub-group comparisons in health-related quality of life if left unadjusted.