The Functional Independence Measurement (FIM) is a new functional status instrument for use among rehabilitation inpatients, but its validity and reliability have been only partially established. Because of its rapid dissemination, we sought further evidence concerning the FIM's internal consistency, responsiveness over time, and construct validity. We examined Uniform Data System (UDS) data on 11,102 general rehabilitation inpatients from the Pacific Northwest. Mean age was 65 and 51% were male. The most common diagnoses were stroke (52%), orthopedic conditions (10%), and brain injury (10%). Internal consistency of the FIM was calculated using Cronbach's alpha. To assess FIM responsiveness, we examined differences between admission and discharge FIM scores. For construct validation purposes, we hypothesized that the FIM would vary with age, comorbidity, discharge destination, and impairment severity. Comorbidity was quantified with the Charlson Comorbidity Index. The FIM had a high overall internal consistency (discharge FIM alpha = .93). The FIM registered significant functional gains during rehabilitation (33% FIM score improvement, p < .001), as do many other functional status indicators. The greatest and least functional improvements were observed for traumatic brain injury and low back pain (53% and 8% FIM score improvement, respectively). The FIM discriminates patients on the basis of age, comorbidity, and discharge destination. Severity differences could be distinguished among spinal cord injury and stroke patients. We conclude that the FIM has high internal consistency and adequate discriminative capabilities for rehabilitation patients. It is a good indicator of burden of care, and demonstrates some responsiveness, but its capacity to measure change over time needs further examination and comparison with competing scales.