Fertility and sibship size of 2,518 psychiatric inpatients during 1968-1975 were compared with national census data and examined according to psychiatric diagnoses, psychiatric diseases in first-degree relatives, early psychic disturbances, duration of disease, and school achievement. Fertility is markedly reduced in all diagnostic subgroups, though particularly in schizophrenia. While psychic disturbances before the age of 15, as an index of a severe disturbance of personality, reduced fertility even further, no significant correlation was found with the duration of the patients' actual disease. Less than ordinary schooling, but also higher qualifications led to a further reduction in fertility. This is particularly so in schizophrenia. Schizophrenics have their children at a later stage of their reproductive career. The psychological and biological consequences of this fact for the transmission of schizophrenia are discussed in detail. No change of fertility relative to the normal population was detected during the observed 8 years. There is suggestive evidence that patients stem from families smaller than expected from the national data. The inherent methodological problems are discussed. The results do not favour the hypothesis of a balanced polymorphism as a mechanism which could explain the constant incidence of psychoses in spite of the severe selection pressure against them.