Background Suicide reduction is government strategyinmanycountries. Weneed to quantify the connection betweennon-fatal self-poisoningand eventual suicide. Aims To determinemortality after an episode of self-poisoningand to identify predictors of death by anycause or by suicide

Declaration of interest None. Fundingdetailed in Acknowledgements. Suicide reduction has recently become the aim of governmental strategies in, for example, England (Department of Health, 1993, 2002), the USA (Vastag, 2001), Australasia and Nordic countries (Taylor et al, 1997), and global strategies through a campaign of the World Health Organization (2004). The main suggested targets for intervention have been people with undetected depression and those recently or currently in psychiatric care – usually with ‘severe mental illness’ (as defined in the various policy documents). Less attention has been paid to those who are known to hospital services because of non-fatal self-harm, although recent primary research (Jenkins et al, 2002; Hawton et al, 2003) and systematic review (Owens et al, 2002) have confirmed that their rate of subsequent suicide is far higher than expected. Unfortunately, much research on suicide after non-fatal self-harm is poor, based on small and highly selected samples, weak methods for detecting suicides during follow-up, and flawed analysis (Owens et al, 2002). We therefore set out to determine long-term mortality and cause of death for around 1000 consecutive patients who attended one of the UK’s largest accident and emergency departments because of selfpoisoning during a brief period in the mid-1980s.