Purpose Measures of quality of life (QoL) are increasingly used to monitor psychological and social well being of clients with chronic illnesses. They are particularly useful in cases where an illness results in multiple disabilities. Where clients have difficulty completing the measure, others who know them may act as proxies. It is important, therefore to assess the accuracy of proxy scores. This study compared the QoL scores given by people with multiple sclerosis (PWMS) with proxy scores given by close relatives. Method The Functional Assessment of Multiple Sclerosis (FAMS) quality of life instrument (Cella et al.) [1] was used. Forty PWMS completed the measure in face-to-face interviews with a speech and language therapist. Close relatives were asked to independently complete the measure on behalf of these clients. Results The overall correlation between client and proxy scores was 0.62, which Sneeuw et al. [2] suggest indicates good agreement. As in previous studies with other groups of clients, proxies underestimated the scores of PWMS. Despite these findings, the results showed substantial disagreements for individual clients and their proxies and a relationship was found which suggests that proxies underestimate the scores of clients with better QoL and overestimate those with poorer QoL scores. Conclusion The level of agreement found between client and proxy scores and the tendency for proxies to underestimate clients' scores are consistent with research with other client groups. The overestimation by proxies of clients who rate their QoL as poor needs further investigation. If confirmed it may complicate the estimation of client scores from those of proxies. We conclude that while proxy scores may be useful as a means of estimating the QoL of groups of clients in research studies they should be used cautiously in the prediction of the QoL of individual clients at the clinical level. Multiple Sclerosis 2007; 13: 1190—1194. http://msj.sagepub.com