Aims. To plan an appropriate response to heroin use in Australia, good esti
mates are needed of the numbers of dependent heroin users, the group who ar
e most in need of treatment, most at risk of fatal opioid overdose and most
at risk of contracting and transmitting blood-borne viruses. Methods. Back
-projection methods were used to estimate the numbers of people starting de
pendent heroin injecting in Australia between 1960 and 1997. Separate analy
ses were based on national opioid overdose deaths and numbers of new entran
ts to methadone treatment in New South Wales (NSW). Estimates of the rates
at which dependent heroin users cease heroin use, commence methadone treatm
ent or die from opioid overdoses were estimated from external data sources.
Results. Back-projection estimates derived from opioid overdose deaths ind
icated that there were 104 000 (lower limit of 72 000 and upper limit of 15
7 000) people who were heroin dependent in Australia between 1960 and 1997.
Of these it was estimated that 67 000 (39 000-120 000) were still heroin d
ependent at the end of 1997. Back-projection estimates based on numbers of
new entrants to methadone treatment in NSW indicated that there were 108 00
0 (82 000-141 000) heroin-dependent people in Australia between 1960 and 19
97, of whom 71 000 (47 000-109 000) were estimated to be heroin dependent a
t the end of 1997. Both analyses indicated that the number of heroin-depend
ent people in Australia has increased substantially from the early 1970s on
wards. Conclusions. Back-projection estimates based on analyses of treatmen
t entries and opioid overdose deaths provide an additional method for estim
ating the numbers of heroin-dependent people in the population. The additio
n of these methods to existing methods, using different data sources and st
atistical methods, should improve consensus estimates of the numbers of her
oin-dependent people.