To compare two methods of heroin withdrawal, 51 heroin users were randomised to undergo a 1 day precipitated withdrawal procedure using naloxone under anaesthetic. About 50 participants were randomised to receive the current standard inpatient withdrawal treatment using clonidine plus symptomatic medication. Following withdrawal, both groups were offered 9 months of naltrexone treatment and supportive counselling. Outcome measures were: commencement of naltrexone, retention in treatment and heroin use at 6 and 12 months. Significantly more of the precipitated withdrawal group completed withdrawal, commenced naltrexone and stayed in treatment for the first 3 months. Overall, there was a significant reduction in both self-reported heroin use and morphine concentration in hair over the 12 month study period, with participants in the precipitated withdrawal group showing significantly lower morphine concentration at 6 months. Being younger and having a lower level of dependence were predictors of abstinence at 6 and 12 months. The advantage of precipitated withdrawal under anesthesia did not persist beyond 3 months with respect to retention in naltrexone treatment or beyond 6 months with respect to heroin use. Long-term follow-up is crucial in assessing the effects of treatment interventions for heroin dependence.