Until the last decade there was considerable confusion about the nosology of childhood psychoses. As Kanner (1958) has pointed out, certain psychodynamically orientated writers (Szurek, 1956; Beres, 1956) have eschewed the important operation of differential diagnosis. This has led to the notion of ‘equality of schizophrenias' (Darr and Worden, 1951) and thus to the idea of a single psychosis of childhood. The controversy over this approach has now waned in the face of empirical evidence from aetiological, phenomenological and follow-up studies, and many authors have stressed the importance of age of onset in their typologies or in their attempts at a more comprehensive classification (Kanner and Lesser, 1958; Mahler et al., 1949, 1952; Bender, 1947, 1959; Anthony, 1958, 1962, and Eisenberg, 1967). As this is also central to the present study it merits examination in greater detail.