This monograph (No. 03) reports on work that tested new methods for estimating the prevalence of problematic heroin use in Melbourne. Using the non-fatal heroin overdose data, three different capture-recapture methods were employed. Estimates were derived for the year 2000 and the year 2003/2004. The lack of plausibility of some estimates coupled with the poor ‘goodness-of-fit’ of some models points to the need to continue to develop new methods for estimating problematic heroin use.