- Background: Understanding whether there is an agreement between older persons who provide information on their functional status and clinicians who assess their function is an important step in the process of creating sound outcome instruments. Objectives: To examine whether there is agreement between self-reported and clinician assessment of similar performance items in older adults. Methods: Fifty independent older adults aged 70-91 years (mean age 80.3 +/- 5.2 years) who live in the community were examined separately and blindly in two data collection sessions. Self-reported and observed lower and upper extremity physical tasks were compared. Life Function and Disability Instrument (LLFDI) was used in both sessions. We performed intra-class correlation coefficients (ICC) as indices of agreement and "mountain plots'' that were based on a cumulative distribution curve. Associations between self-reported and observed function with Fear of Fall Scale (FES) and Geriatric Depression Scale (GDS) were also assessed. Results: ICCs were high between self-reported lower extremity function and observed lower extremity function (ICC = 0.83), and were poorer for self-reported and observed upper extremity function (ICC = 0.31). In both comparisons, mountain plots revealed a right shift that was larger for upper than lower extremity functions, indicating systematic differences in self-reported and observed assessments. Associations with FES and GDS were higher for self-reported than observed function. Conclusion: There is a systematic bias between self-reported and clinician observation. Professionals should be aware that information provided by patients and observation of activity assessed by clinicians could differ substantially, especially for upper extremity function.