Variation in their year and month of birth was studied in the 13,661 schizophrenics born between 1914 and 1960 known to the Scottish Psychiatric Case Register. Year-to-year fluctuations in the number of schizophrenics per 10,000 live births were outside chance limits. So were month-to-month fluctuations between January 1932 and December 1960, and this was largely due to fluctuations in the numbers of schizophrenics born in February, March, April and May. Time-lagged correlations with mean monthly temperatures suggest that in these same four months the incidence of schizophrenia is influenced by temperature six months previously - the lower the temperature in the autumn the higher the incidence of schizophrenic births the following spring. If these findings can be confirmed in other data sets, they would suggest that some influence which varies consistently with season and temperature is contributing to the aetiology of schizophrenia and may exert its effects as early as the third or fourth month of foetal development.