We extended research originating with the Tsuang-Winokur criteria for paranoid and nonparanoid schizophrenia. To partially resolve problems of subtype instability, 41 consecutive admissions meeting DSM-III-R criteria for schizophrenia were subdivided according to whether they ever experienced prominent systematized delusions. Neuropsychological profiles for paranoid patients and nonparanoids with a history of systematized delusions were extremely similar. When combined, this 'systematized' group had significantly better general verbal ability and verbal memory than patients who never manifested systematized delusions. There was also a significant neuropsychological function-by-group interaction. The neuropsychological data suggested that systematized patients had better premorbid cognitive functioning as well as a greater discrepancy between premorbid verbal ability and current attentional functioning. No between-group differences were found on a measure of prefrontal-executive function, nor were there any neuropsychological differences between traditionally defined (DSM-III-R) paranoid and nonparanoid subgroups. This study suggests a possible shift in the dividing line between paranoid and nonparanoid subtypes and illustrates the potential value of neuropsychological data for refining psychiatric nosologies.