The health-related quality of life (HRQL) of people with dementia has received growing attention in formulating decisions about the provision and financing of health and social care. There is a need for measurement perspectives to determine whether HRQL assessment has captured what is important to the target population, to generate a coherent body of evidence to guide clinical and policy decisions. The thesis first investigates if HRQL in dementia is meaningfully interpreted as a general phenomenon in which the whole is greater than the sum of its parts. Overall total scores on DEMQOL and DEMQOL-Proxy were more sensitive to a general theme of individual differences in HRQL than subscale scores from multiple themes. Next, based on this measurement perspective, inconsistencies in self- and informant report behaviour were examined between geographical region, gender, and dementia severity. Items that demonstrated desirable measurement properties at this stage were selected for the short-form versions, DEMQOL-SF and DEMQOL-Proxy-SF, which included the preferencebased items in DEMQOL-U and DEMQOL-Proxy-U. These provided the basis for a set of analyses exploring whether changes in subjective HRQL are influenced by response shift in meaning, priorities, or expectations over time. The thesis reports the findings that differences that emerge over repeated HRQL assessments could not be attributed to re-conceptualisation, re-prioritisation, or recalibration of internal standards. Furthermore, differences in raw total scores over time were sensitive to HRQL improvement or deterioration. As such gains or losses in utility values from DEMQOL-U and DEMQOL-Proxy-U items would be consistent with item responses that reflect longitudinal changes in HRQL. Taken together, the thesis suggests that the DEMQOL measurement system has tenable foundations for the clinical and economic evaluation of HRQL changes in dementia treatment interventions across clinical and social care settings.