The data presented by the authors suggest a more successful prediction of outcome by the process-reactive dimension when it is applied to schizophrenic patients defined according to broad concepts of schizophrenia (e.g., DSM-II) than to narrow concepts of schizophrenia (e.g., DSM-III). Research indicates that the DSM-III is more likely than the DSM-II diagnosis to be composed of "poor prognosis" schizophrenia, as defined by the classical prognostic indicators. In general, the overall results could support the views of Herron and others that the process-reactive dimension is more valuable as a predictor for a schizophrenia spectrum than for modern, narrow concepts of schizophrenia. However, while recent data of the authors indicate that DSM-III schizophrenic patients show poorer outcome than DSM-II schizophrenic patients, they also indicate that (1) DSM-III schizophrenic patients still show heterogeneity in prehospital functioning and outcome, and (2) other prognostic indexes, such as the Zigler-Phillips Scale and the Vaillant-Stephens Scale, have some efficacy in predicting course and outcome among DSM-III schizophrenic patients. Data on prehospital functioning and outcome are discussed in terms of two contrasting models--a developmental model and a natural history model of the course of schizophrenia.