Various methods of subtyping schizophrenia are briefly reviewed. It is concluded that the paranoid-nonparanoid dichotomy has been the most consistently supported both clinically and in experimental research. In an attempt to deal with its limitations, a tentative reconceptualization is proposed for schizophrenia. This new model is based on two independent factors: severity of disorder and severity of symptom. The present dimensional approach is shown to be consistent with a number of empirical findings regarding differences and associations between paranoid and nonparanoid symptomatology. The relative advantages of dimensional and categorical descriptions, as well as their combination, are presented.