The paper sets out the arguments in favour of utilising both a 'strict' definition corresponding to the 'nuclear' concept of schizophrenia and a 'broader' definition that also subsumes 'atypical' schizophreniform and related syndromes. Evidence is adduced rejecting the view that attempts to effect a complete separation between 'nuclear' and other forms of schizophrenic disorders based on phenomenological, hereditary, prognostic, etiological and other lines. Both the unity and the diversity of the schizophrenic family of disorders must be given explicit recognition in clinical and scientific work. The value of the 'nuclear' concept has been demonstrated by the IPSS study among other enquiries. The advantages of the broad definition are twofold. It makes for a flexibility and openness of approach in diagnosis and treatment of schizophrenia that are indespensable at the present time. The value of the 'broad' definition is that it encompasses a range of phenomena that have been shown to be related to one another. They may therefore be compared and contrasted. It is argued that new knowledge about the etiology of schizophrenia may be acquired in this manner. Tentative definitions for the 'nuclear' and 'broad' groups of disorder are given.