Confirmatory factor analysis was used to examine a proposed factor structure of a comprehensive neuropsychological battery used to study patients with schizophrenia and related psychotic disorders (n = 209). An a priori six-factor model and five nested models were evaluated successively, using maximum likelihood confirmatory factor analysis. In all multifactor models, the factors were significantly intercorrelated. A six-factor model with two pairs of correlated errors fit the neuropsychological data significantly better than competing models with fewer factors. The six factors included verbal crystallized, attention/working memory, verbal episodic memory, speed of information processing, visual episodic memory, and reasoning/problem solving. Severity of negative symptoms was significantly associated with worse performance on attention/working memory and verbal crystallized factors, but positive symptoms, depression, and a summary measure of psychopathology were not significantly related to neuropsychological performance. Impairment on a performance-based measure of functional capacity was significantly related to all neuropsychological factors. A simultaneous confirmatory factor analysis using the original sample and a group of healthy subjects (n = 131) demonstrated that the six-factor model of cognition was generalizable and applied equally well to both groups.