The latent structure of the California Verbal Learning Test-Second Edition (CVLT-II) was examined in a clinical sample of 223 persons with traumatic brain injury that had been screened to remove individuals with complicating premorbid (e.g., psychiatric) or comorbid (e.g., financial compensation seeking) histories. Analyses incorporated the "z" scores from 12 CVLT-II variables. Maximum-likelihood confirmatory analyses were performed to test the fit and parsimony of four hypothetical models. A four-factor model, consisting of Attention Span, Learning Efficiency, Delayed Memory, and Inaccurate Memory, met all the a priori specified criteria for model fit and parsimony. This model was consistent with that identified previously in a confirmatory factor analysis of the CVLT-II standardization sample. The results support the construct validity of the CVLT-II in individuals with traumatic brain injury and indicate that a multifactorial interpretation is appropriate for clinical practice. (Contains 4 tables.)