The computerized medical file, used in routine work in an Adult Psychiatry University-Hospital Unit enabled us to select 113 cases among 1,000 consecutive hospitalizations, the diagnosis of which could possibly lead to schizophrenia. These cases which we named "paraschizophrenic states" are linked to DSM III-R criteria of borderline (27 cases), schizoid (40 cases) or schizotypical (15 cases) personalities, schizophreniform trouble and unspecified psychotic trouble (17 cases), brief reactional psychosis (14 cases). We selected 196 cases of schizophrenia in the same cohort of hospitalized patients. As it is now usually admitted, we marked out two subgroups in this second group: the positive schizophrenia which gather together the paranoid and undifferentiated patterns and the negative schizophrenia which correspond to disorganized, catatonic and residual models, according to DSM III-R criterion. We compared the "paraschizophrenic states'" group and its five subgroups (we indeed joined schizophreniform trouble and unspecified psychotic trouble under the name of "other psychotic trouble" by reason of their relative nosographic lacks of precision and of their too small sizes) with the schizophrenia's group and its two subgroups. Each group is matched for sex (1.51 men for 1 woman in the first group and 1.45 men for 1 woman in the second group). We evaluated statistics for markers usually studied in schizophrenia in each subgroup. These markers are of three classes: biographical: age during the study, age of troubles' onset, season of birth; socioeconomic: socioeconomic level of family and patient's student status; psychiatric: family (history affective trouble, psychotic trouble, alcoholism), treatment response and short- and middle-term prognosis.