BACKGROUND & AIMS: Heroin use is associated with a disproportionately high level of morbidity and mortality with most deaths attributable to drug overdose. Aggregate heroin purity data has been used to examine the relationship between overdose and variability in street-level heroin, however heroin purity data alone may not be the most appropriate nor a sensitive enough measurement tool for this assessment. The aim of this study was to measure the variability in effective dose of street-level heroin seizures, accounting for variation in both purity and mass, and determine the proportion of samples with higher than expected effective dose that would not be detected using a purity-only measure. METHODS: Data on Victorian heroin seizures </=150mg in mass made between 01/01/2012 and 31/12/2013 were obtained from the Victoria Police Forensic Services Department. The effective dose of heroin in each sample was determined by multiplying the mass and purity variables. Effective dose outlier samples were considered as those containing either greater than 1.5-2 times or >2 times the median effective dose of heroin for the sample data. RESULTS: The 983 street-level heroin samples of </=150mg had a median mass of 92mg (IQR of 43mg), a median purity of 13% (range 3.6%-80.9%) and a median effective dose of 12.0mg of heroin (IQR 6.6mg; range 0.4mg-111mg). Approximately one in 13 samples (8%) and one in 17 samples (6%) contained between 1.5-2 times and >2 times the median effective dose of heroin respectively. CONCLUSION: The effective dose of heroin is a more appropriate measure than purity to identify outlier samples that containing larger than expected doses of heroin compared to typical doses that may be expected by users. Together with other identified risk factors, fluctuation in the effective dose of heroin contained in street-level samples may contribute to the potential for overdose.