The usefulness and validity of traditional subtypes are questionable. The subtypes described in earlier years no longer emerge with the clarity previously described. The four classical subtypes cannot be reliably distinguished and have not been shown to have predictive validity. Subtypes classified along course or prognostic lines may be more clinically useful. Attempts to subdivide schizophrenia along biologic and genetic lines offer promise. Recent efforts to describe new subdivisions of schizophrenia are readily justified, but new descriptive subtypes are likely to prove useful only when validated by biological, genetic, treatment response, and outcome data.