The functional independence measure (FIM) is used to determine the degree of disability that patients experience and the progress that they make through programs of medical rehabilitation. Rasch analysis is a statistical technique for constructing interval measures from ordinal data that was applied to derive FIM measures. The major factors that are taken into account to produce FIM measures are the relative difficulty in performance of FIM items and the ability of the persons tested. Our analyses showed the relative difficulties that patients experienced in performing items in the FIM. There were two dominant patterns of difficulty, one for motor FIM items and the other for cognitive FIM items. The patterns were consistent across impairment groups, although not identical. Of the motor items, eating and grooming were easiest whereas stair climbing, tub/shower transfers and locomotion were most difficult. Of the cognitive items, expression and comprehension were easiest and problem solving was the most difficult. The patterns of difficulty in performing FIM items are illustrated by analysis of the following impairment groups: for motor items, orthopedic conditions, stroke with left hemiparesis and spinal cord dysfunction; for cognitive items, orthopedic conditions, brain dysfunction, stroke with right hemiparesis and spinal cord dysfunction. By understanding patterns of difficulty in performing FIM items according to types of impairment and levels of function, clinicians may more precisely design treatment programs, use services and predict outcomes of medical rehabilitation.