OBJECTIVE: To explore how clinical and demographic variables impact on the management of diabetes mellitus in general practice. DESIGN: A structured vignette survey was conducted in Australia. This included nine vignettes chosen at random from 128 developed around seven clinical variables. Respondents were asked to recommend a change in treatment and make specific recommendations. A random sample of general practitioners (GPs) were recruited. Two diabetologists involved in the development of national guidelines also participated. RESULTS: 125 (13.8%) GPs participated. Statistical analyses were used to generate outcome measures. GPs recommended a change in treatment for most (81.1%) cases; were less likely to prescribe a statin (68.5% GPs vs. 76.3% diabetologists), less likely to treat hypertension (66.7% vs.89%) and less likely to refer for lifestyle modification (82.3% vs. 96.5%). Significant disagreement occurred around prescribing or changing oral hypoglycaemics. No GP characteristics showed significant impact. The proportion of GPs who agreed with diabetiologists on dose and choice of drugs was 35.7% for statins, 49.6% for antihypertensives and 39.6% for oral hypoglycaemics. CONCLUSIONS: There were significant differences between diabetologists and GPs on the management of diabetes. The survey suggests significant under-dosing by GPs. These findings warrant further investigation.