OBJECTIVES: Type 1 diabetes mellitus (T1DM), resulting from an immune-associated destruction of insulin-secreting pancreatic β-cells, has been reported in a few earlier studies to be inversely associated with schizophrenia, but not with schizophrenia-like psychoses. The aim of this study was to verify this finding by carrying out a Swedish register study. METHODS: Data from the Total Population- and Medical Birth-Registers were used to create a cohort of all individuals born in Sweden 1987-2004. The cohort individuals were linked with the Inpatient- and Outpatient-Registers and followed from birth to 2017 to identify onset of T1DM, schizophrenia and schizoaffective disorder. Cox proportional hazard regression models were used to assess the association between T1DM and risk of developing schizophrenia or schizoaffective disorder during a follow-up from age 13. RESULTS: The study population included 1 745 977 individuals and the length of follow-up was maximally 18.0 (median 9.7) years. During the follow-up, 1 280 individuals developed schizophrenia and 649 individuals schizoaffective disorder. The risk of developing schizophrenia was significantly lower among individuals with, than among individuals without, a diagnosis of T1DM, whereas the risk of developing schizoaffective disorder did not differ among individuals with or without a T1DM diagnosis [adjusted hazard ratio (95% confidence interval); schizophrenia: 0.29 (0.09-0.91), p=0.0338, schizoaffective disorder: 1.50 (0.71-3.16), p=0.2909]. CONCLUSIONS: This study, in line with previous studies, shows that a diagnosis of T1DM is associated with a decreased risk of schizophrenia. This finding of an inverse association between T1DM and schizophrenia may bring an interesting piece, related to autoimmunity, into the schizophrenia-aetiology puzzle.