Pregnant injecting drug users were randomly assigned to: (i) individually receive a six-session cognitive-behavioural intervention in addition to their usual methadone maintenance treatment (intervention condition (I) (n = 40)); or (ii) their usual methadone maintenance treatment only (control condition (C) (n = 40)). There was no change in drug use per se in either group after the intervention. However, at 9-month follow-up the I group had significantly reduced some HIV risk-taking behaviours (in particular injecting risk behaviours). The I group reduced the needle risk associated both with "typical" use (drug use in the month before interview) and "binge" use (drug use in the month nominated as the heaviest month of drug use in the previous 6 months). The intervention had no effect on sexual risk behaviours. The finding of reduced injecting risk behaviour following the six-session intervention suggests that such an intervention may be of benefit for individuals persisting with injecting risk behaviours despite methadone maintenance treatment and the availability of sterile injection equipment.