In reply: We disagree with Wodak and colleagues in a number of respects.

• If the results of a hydromorphone trial were as good as Wodak et al claim a heroin trial would be, then a heroin trial would be unnecessary.

• We accept, as did the Dutch and Swiss, that politicians have the authority to make decisions about what medical research is permitted. The heroin trials in Switzerland and the Netherlands were approved by parliament and supported by referenda in cantons and cities in Switzerland.

• The Australian survey data cited by Wodak and colleagues indicate that a heroin trial would not have been approved if a referendum had been held in 2001. Nor do we think it would be supported by a free vote in Federal Parliament, as it was not supported by a similar vote at the NSW Drug Summit in 1999.

• Wodak et al present no evidence to support their claim that the delivery of a treatment that costs between A$25 000 and A$45 000 per patient per year in the Netherlands1 to less than 5% of the heroin-dependent population would have a detectable impact at the population level.

A hydromorphone trial would provide a way of evaluating the role of injectable opioids in the treatment of heroin dependence. It would not prevent Wodak and colleagues from convincing the community that injectable heroin is the drug of choice to treat refractory heroin dependence.