Objective: To empirically compare the measurement properties of the DEMQOL-U and DEMQOL-Proxy-U instruments to the EQ-5D-5L and its proxy version (CEQ-5D-5L) in a population of frail older people living in residential aged care in the post-hospitalisation period following a hip fracture. Methods: A battery of instruments to measure health-related quality of life (HRQoL), cognition, and clinical indicators of depression, pain and functioning were administered at baseline and repeated at 4 weeks’ follow-up. Descriptive summary statistics were produced and psychometric analyses were conducted to assess the levels of agreement, convergent validity and known group validity between clinical indicators and HRQoL measures. Results: There was a large divergence in mean (SD) utility scores at baseline for the EQ-5D-5L and DEMQOL-U [EQ-5D-5L mean 0.21 (0.19); DEMQOL-U mean 0.79 (0.14)]. At 4 weeks’ follow-up, there was a marked improvement in EQ-5D-5L scores whereas DEMQOL-U scores had deteriorated. [EQ-5D-5L mean 0.45 (0.38); DEMQOL-U mean 0.58 (0.38)]. The EQ-5D and CEQ-5D-5L were more responsive to the physical recovery trajectory experienced by frail older people following surgery to repair a fractured hip, whereas the DEMQOL-U and DEMQOL-Proxy-U appeared more responsive to the changes in delirium and dementia symptoms often experienced by frail older people in this period. Conclusions: This study presents important insights into the HRQoL of a relatively under-researched population of post-hospitalisation frail older people in residential care. Further research should investigate the implications for economic evaluation of self-complete versus proxy assessment of HRQoL and the choice of preference-based instrument for the measurement and valuation of HRQoL in older people exhibiting cognitive decline, dementia and other co-morbidities.