The ability of neuropsychological tests to predict rehabilitation outcome is unclear, particularly when other ratings of cognition are available. Neuropsychological test scores and functional ratings of cognition (Functional Independence Measure (FIM) Cognition score) were used to predict improvement in patient mobility and self-care skill, as measured by the FIM Motor score. Regression models used both raw neuropsychology test scores and age-adjusted scores. Retrospective chart review was performed for patients on an inpatient rehabilitation unit and referred for neuropsychological assessment. The group included 126 subjects (average age 64.2 ± 17.1 years) and a variety of medical diagnoses. Neuropsychological tests included the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS). After forcing the Admission FIM Cognition score into the model, RBANS scores and duration of rehabilitation predicted FIM Motor improvements (F = 11.42, p < 0.0001). Raw neuropsychological test scores performed better than the model with age-adjusted test scores. FIM Cognition alone did not predict FIM Motor improvements. Neuropsychological tests, combined with duration of rehabilitation, predicted mobility gains for patients undergoing inpatient rehabilitation beyond what was predicted by another, readily available, assessment of cognition. Neuropsychology raw scores performed better than age-adjusted scores, raising questions about the standard use of demographic adjustments for predicting real-world function.