This article critiques recent studies of the effectiveness of inpatient and outpatient rehabilitation after traumatic brain injury with respect to how well they have controlled for the nature and severity of initial brain injury, time since injury, and preinjury characteristics of the subjects. These variables potentially confound the results of studies attempting to determine whether rehabilitation after traumatic brain injury results in improved outcomes and whether it is cost effective. To advance this field of research, investigators must be more uniform and explicit in the characterization of their subjects with respect to severity of injury and preinjury characteristics and more uniform in their selection of functional outcome measures, and they must use control groups of patients who receive few services or alternative treatments