As part of the evaluation of a drink driver rehabilitation program in a non-metropolitan area of Queensland we collected retrospective and prospective information on close to 900 drink drivers who undertook the program and over 800 contemporaneous drink driving offenders from the same area who did not. The data collected included details of the index drink driving offence, traffic and criminal histories for 5 years prior to the index court hearing and subsequent criminal and drink driving offences. Over 200 drink driving re-offences were committed by the cohort during the follow-up period, which averaged 34 months, with a range of 15 to 57 months. Using a failure-time model of time to re-offence, including proportional hazards modelling, independent predictors of re-offence among program participants and were identified. These include previous drink driving, unlicensed or dangerous driving, and a history of criminal convictions of any type.