<b><i>Background:</i></b> The aim of the study was to estimate the lethality of opioid overdose among young heroin users. <b><i>Methods:</i></b> A prospective community cohort study was conducted in Barcelona and Madrid, Spain. Participants included 791 heroin users aged 18-30 years who were followed up between 2001 and 2006. Fatal overdoses were identified by record linkage of the cohort with the general mortality register, while non-fatal overdoses were self-reported at baseline and follow-up interviews. The person-years (py) at risk were computed for each participant. Fatal and non-fatal overdose rates were estimated by city. Transition towards injection shortly before the overdose could not be measured. Overdose lethality (rate of fatal overdose in proportion to total overdose) and its 95% CI was estimated using Bayesian models. <b><i>Results:</i></b> The adjusted rates of fatal and non-fatal opioid overdose were 0.7/100 py (95% CI: 0.4-1.1) and 15.8/100 py (95% CI: 14.3-17.6), respectively. The adjusted lethality was 4.2% (95% CI: 2.5-6.5). <b><i>Conclusions: </i></b>Four out of 100 opioid overdoses are fatal. These are preventable deaths that could be avoided before or after the overdose takes place. Resources are urgently needed to prevent fatal opioid overdose.