As the use of proxy respondents is sometimes necessary in research involving older or disabled people it is important to assess the impact of this on research results. This study examined the concordance of proxy responses and index responses on questions concerning functional abilities and continence. Index respondents were women aged 65 years and older who were interviewed as part of a larger study of outcomes of fractured neck of femur, at 5 days post surgery. They were asked to nominate a proxy respondent who was interviewed using the same questions within a few days. Responses of proxy and index respondents were compared using percentage agreement and Kappa statistics. On questions regarding functional ability, proxy responses were found to be more reliable for personal care activities than for instrumental activities of daily living. This may be a result of questions concerning instrumental activities being somewhat more ambiguous and open to individual interpretation. Item non-response was low for all questions thus resulting in little missing data for proxy respondents. There was a tendency for more distant relationship and contact to produce better agreement, which is contrary to previous findings. Proxy responses were biased in the direction of an overestimation of functional incapacity and so researchers should be cautious in combining data from proxy and index respondents. Concordance was good for questions concerning urinary and faecal incontinence although non-response was higher than for functional ability questions. Concordance was not as great for more detailed questions concerning the timing and frequency of incontinence as these used graded response options, rather than simple yes/no responses.