To fill a gap in knowledge about the effectiveness of brief intervention for hazardous alcohol use among Indigenous Australians, we attempted to implement a randomised controlled trial in an urban Aboriginal Medical Service (AMS) as a joint AMS-university partnership. Because of low numbers of potential participants being screened, the RCT was abandoned in favour of a two-part "demonstration project". Only 16 clients were recruited for follow-up in six-months, and the trial was terminated. Clinic, patient, Aboriginal health worker, and GP factors, interacting with study design factors, all contributed to our inability to implement the trial as designed. The key points to emerge from the study are that alcohol misuse is a difficult issue to manage in an Indigenous primary health care setting; RCTs involving inevitably complex study protocols may not be acceptable or sufficiently adaptable to make them viable in busy, Indigenous primary health care settings; and "gold-standard" RCT-derived evidence for the effectiveness of many public health interventions in Indigenous primary health care settings may never be available, and decisions about appropriate interventions will often have to be based on qualitative assessment of appropriateness and evidence from other populations and other settings.