Choosing between preference-based instruments of health-related quality of life (HR-QOL) in particular situations is an important area for research. Even where instruments can be assumed to be measuring the same thing, they may not be interchangeable. The study presented investigates the extent to which EQ-5D and SF-6D instruments are interchangeable in an older, chronically ill patient group undergoing haemodialysis. Head-to-head comparisons were made using 'practicality', 'descriptive validity', 'empirical validity', mean utilities and associated distributions. Overall it was difficult to choose between instruments on the basis of descriptive or empirical validity, since both performed similarly. Important differences were, however, found relating to practicality: a significantly higher response rate in favour of EQ-5D; and lower levels of missing data to derive health states. Non-response was significantly associated with age and co-morbidity of respondents. We suggest that in patients undergoing haemodialysis, and potentially other older chronically ill patient groups, EQ-5D is the primary preference-based generic HR-QOL instrument.