Objective : The present study further investigated the factor structure of the WCST in traumatic brain injury and investigated the construct validity and relationships among scores through the use of cluster analysis. Design : Participants were 68 survivors of chronic severe TBI, living at a residential brain injury rehabilitation facility. Methods and procedures : Three sets of WCST scores were submitted to factor analysis; the regression factor scores based on the standard WCST were examined using cluster analysis. Results : Factor analysis of the WCST raw scores replicated the three-factor solution which has been previously reported. When t -scores were analysed, two-to-four-factor solutions could be justified. The cluster analysis identified four groups representing: (1) impaired response maintenance; (2) problem-solving deficits; (3) intact WCST performance; and (4) deficits in set shifting. Conclusions : The results support previous research indicating that the WCST is sensitive to three distinct cognitive processes: cognitive flexibility, problem-solving, and response maintenance. However, unlike the cognitive processes underlying WCST performance, the WCST scores representing these processes are not independent. The potential clinical relevance of these results is discussed.