Fatal overdose and drug-related mortality are key harms associated with heroin use, especially injecting drug use (IDU), and are a significant contribution to premature mortality among young adults. Routine mortality statistics tend to underreport the number of overdose deaths and do not reflect the wider causes of death associated with heroin use. Cohort studies could provide evidence for interpreting trends in routine mortality statistics and monitoring the effectiveness of strategies that aim to reduce drug-related deaths. We aimed to conduct a retrospective mortality cohort study of heroin users recruited from an anonymous reporting system from specialist drug clinics. Our focus was to test whether (1). specialist agencies would agree to participate with a mortality cohort study, (2). a sample could be recruited to achieve credible estimates of the mortality rate, and (3). ethical considerations could be met. In total, 881 heroin users were recruited from 15 specialist drug agencies. The overall mortality rate of the cohort of heroin users was 1.6 (95% confidence interval [CI], 1.1 to 2.2) per 100 person-years. Mortality was higher among males, heroin users older than 30 years, and injectors, but not significantly higher after adjustment in a Cox proportional hazard model. Among the 33 deaths, 17 (52%) were certified from a heroin/methadone or opiate overdose, 4 (12%) from drug misuse, 4 (12%) unascertained, and 8 (24%) unrelated to acute toxic effects of drug use. Overall, the overdose mortality rate was estimated to be at least 1.0 per 100 person-years. The standardized mortality ratio (SMR) was 17 times higher for female and male heroin users in the cohort compared to mortality in the non-heroin-using London population aged 15-59 years. The pilot study showed that these studies are feasible and ethical, and that specialist drug agencies could have a vital role to play in the monitoring of drug-related mortality.