Studies of the risk of schizophrenia and related non-affective psychoses among individuals exposed to head injury have reported conflicting findings. The diagnostic specificity of reported associations remains unclear and the effects of sex, genetic vulnerability and timing of exposures are also uncertain. The authors used a comprehensive range of data derived from the linkage of several Swedish registers to investigate associations of head injury with schizophrenia and other non-affective psychoses. We carried out a nested case control study drawn from a cohort of 785,051 men and women born between 1973 and 1980. After head injury exposure, there was a small increase in the risk (OR 1.37 (CI 1.14-1.66)) of non-schizophrenic, non-affective psychoses but not schizophrenic disorder (OR 1.10 CI 0.82 to 1.47). There was no critical period effect for exposures in early childhood, the association with non-affective psychosis being observed only in those exposed to head injury after age 10. There was no evidence that associations differed in males and females or in those with, and those without, a family history of psychosis. We conclude that it is unlikely that head injury causes schizophrenia. There is a small but significant risk of non-schizophrenic psychotic disorders but this is not associated with early age of injury or family history of psychosis. Further research is needed to clarify the potential roles of social and biological intermediary factors.