Empirical research on the diagnosis of schizophrenia during the late 1970s is reviewed. This period covers work carried out just before the introduction of DSM-III. Two trends are apparent. First, the concept of schizophrenia is narrowing as a result of a decreased emphasis on "classic" schizophrenic symptoms, an increased reliance on affective symptoms as more diagnostically useful, a shift toward classifying schizoaffective disorder as more similar to affective diagnoses, and an increase in the use of explicitly defined operational criteria derived with an emphasis on reliability. Second, the concept of schizophrenia as a unitary disease is diminishing. Heterogeneity is becoming a focus of study, and biological variables may define subtypes. As definition and reliability have been the nosological problems of the 1970s, subtyping and comparative validity among diagnostic systems may be the problems for the 1980s.