When inferring brain dysfunction, test scores are typically compared to normative data based on estimates of premorbid intelligence (e.g., by educational level or reading scores). However, these methods are likely to lead to differing results, with important diagnostic and forensic implications. The current study compared estimates of impairment (reported in z-scores) based on educational level versus reading scores in a population with traumatic brain injury. The study included 174 patients (M age = 27.3; M education = 12.3) evaluated as outpatients at a university hospital rehabilitation department. Wilcoxen ranked sign tests indicated that the two methods yielded estimates that were statistically different (p < .0001) for all variables. The education based method yielded greater estimates of impairment than the reading score method for WAIS-R FIQ, Grip Strength, and Finger Tapping, with a pattern of generally consistent impairment across cognitive/motor areas (z-score range = -0.59 to -.97). In contrast, the reading score based method yielded greater estimates of impairment in processing speed (Trails A) and flexibility (Trails B), with a wider range of impairment noted between cognitive and motor domains (z-score range = +0.21 to -2.95). Clinical implications are discussed.