Background: Currently there is a lack of strong epidemiological evidence for violent offending and intelligence quotient (IQ) level as predictors of suicide risk among people with schizophrenia.Methods: We interlinked several Swedish routinely collected national registers, including the Hospital Discharge, Crime, Conscript, Migration and Cause of Death Registers, to identify 13,804 patients admitted at least twice with a schizophrenia diagnosis during 1973-2004.All deaths by specific cause were ascertained all and suicides identified, and rate ratios were estimated using Poisson regression models with adjustment for age and period effects.Results: There were 871 suicides: 6.3% of the cohort.Almost 80% of these cases were aged 20-39 years, and 80% occurred within 10 years of second discharge, with almost a quarter taking place within a year.The following risk factors were identified: younger age, 15-19 versus 40 years and older (rate ratio 3.18, 95% confidence interval 1.82-5.56), male sex (rate ratio 1.37, 1.18-1.59) and history of violent offending (rate ratio 1.45, 1.21-1.73).Intelligence quotient data were unavailable for women, but in men, lower IQ appeared protective compared to those of average or higher levels of intelligence (rate ratio 0.71, 0.58-0.86).Further stratified analyses to identify effect modification indicated that the elevated risk associated with past violent offending was restricted to males, and specifically to those with a lower IQ.Conclusion: Some factors associated with suicide may be specific to schizophrenia, such as the higher risk seen among teenagers or young adults with the disorder, and the protective effect of having a low IQ.However, we have also shown that history of violent offending is an important risk factor, as is the case in the general population.We found that the effect of violent offending is modified by gender and by IQ level, which may better inform clinical risk management.Researchers now need to identify other factors that modify suicide risk among those schizophrenia patients with past violent offending, and also achieve a better understanding of the causes of these complex relationships.