The concept of "childhood schizophrenia" is ambiguous because of problems on the diagnostic level, especially in regard to symptomatology and prognosis. After studies which produced the description of "early infantile autism", childhood schizophrenia studies were even more confused. One reason for this was that there were not sufficient follow-up studies about the schizophrenia-like psychoses during occurring childhood, on the one hand, and the retrospective-anamnestic research was limited in the field of adult psychiatry, on the other hand. The author has focused on two points of view in order to discuss the problem of childhood schizophrenia. One is the non outbreak of childhood psychoses in children from 6 to 8 years old which was demonstrated by a couple of epidemiological research studies. The other is the concept of "Knick" (in Germany) which usually accompanies classical German theories of schizophrenia. "Knick" means that, at some point in the patient's history, there is an incident that is the indication of the onset of schizophrenia. The subjects were selected from the author's own cases. They continuously and insidiously developed schizophrenic symptoms during pre-puberty and puberty, presenting behavioral abnormalities or neurosis-like signs from early childhood. The diagnostic criteria for schizophrenia corresponds to DSM-III and with the German theory of Bleuler and Schneider. All cases have been treated for more than ten years. The acute onset cases, the mentally retarded cases (DSM-III .318) and the cases with suspected organic brain disorders were excluded because of the diagnostic problems. The author could differentiate these subjects into three groups based on the development in the infantile stage. The first was autistic group and the second mentally subnormal (DSM-III .317) or borderline group. Psychiatric problems of these two groups were already obvious in early childhood. The third group developed normally until they were 3-5 years old. After that they presented neurotic symptoms having some kind of intra-familiar problems. The author could confirm that these three groups manifested various kinds of neurosis-like symptoms, mainly obsessive-compulsive ones, from 3-5 years old. The so-called latent time, which is thought to be a non-outbreak time of childhood psychoses, may correspond to the prodromal stage of schizophrenia in these cases. This research is based on study of 31 children. An in-depth study has been done by the author of 6 of these children, 2 in each of the 3 categories described above. The "Knick"-points were not identifiable in these cases.(ABSTRACT TRUNCATED AT 400 WORDS)