Objectives: To describe the management of cardiovascular disease (CVD) risk in Australian patients with diabetes; to compare the effectiveness of a quality improvement initiative for people with and without diabetes. Research design and methods: Subgroup analyses of patients with and without diabetes participating in a cluster randomised trial. Setting and participants: Indigenous people (≥ 35 years old) and non-Indigenous people (≥ 45 years old) who had attended one of 60 Australian primary health care services at least three times during the preceding 24 months and at least once during the past 6 months. Intervention: Quality improvement initiative comprising pointof- care electronic decision support with audit and feedback tools. Main outcome measures: Adherence to CVD risk screening and prescribing guidelines. Results: Baseline rates of guideline-recommended screening were higher for 8829 patients with diabetes than for 44 335 without diabetes (62.0% v 39.5%; P < 0.001). Baseline rates of guideline-recommended prescribing were greater for patients with diabetes than for other patients at high risk of CVD (55.5% v 39.6%; P < 0.001). The proportions of patients with diabetes not attaining recommended treatment targets for blood pressure, low-density lipoprotein-cholesterol or HbA1c levels who were not prescribed the corresponding therapy at baseline were 28%, 44% and 24% respectively. The intervention was associated with improved screening rates, but the effect was smaller for patients with diabetes than for those without diabetes (rate ratio [RR], 1.14 v 1.28; P = 0.01). It was associated with improved guideline-recommended prescribing only for undertreated individuals at high risk; the effect size was similar for those with and without diabetes (RR, 1.63 v 1.53; P = 0.28). Conclusions: Adherence to CVD risk management guidelines was better for people with diabetes, but there is room for improvement. The intervention was modestly effective in people with diabetes, but further strategies are needed to close evidence–practice gaps.