Neuropsychological assessment is frequently requested for answering questions about real-world functioning. The ecological validity of neuropsychological instruments in making such predictions about older adults remains largely unknown. Only a handful of studies have examined relationships among cognitive and functional abilities and have suggested a moderate degree of relationship between global performance scores on functional and cognitive screening instruments. But relationships between performance on specific cognitive tests and within common functional domains have remained understudied. The relationships among a comprehensive range of neuropsychological tests, including the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS; Randolph, 1998), everyday functional skills measured using the Direct Assessment of Functional Status (DAFS; Loewenstein et al., 1989), measures of prospective memory from the Rivermead Behavioural Memory Test (RBMT; Wilson, Cochburn, & Baddeley, 1991), and self- and caregiver-report of functional memory skill were presently evaluated among assisted-living community residents. RBANS total scores only accounted for slightly more variance (48%) than specific cognitive tests (43%) in prediction of DAFS performance, and superior prediction was made when combining both global and test-level in regression on DAFS (52%). Rate of explained variance improved to between 56% - 61% when prospective memory scores were summed with the criterion measure (IADL). Planned models were also used to illustrate the value of unconventional functional criteria relative to traditional IADL commonly used in the validation of cognitive measures. The present findings did not strongly support the use of report measures for the evaluation of everyday memory decline or functional skill in assisted-living populations.