AIMS: To evaluate basal and prandial insulin initiation and titration in people with type 2 diabetes mellitus (T2DM) in primary care and to explore the feasibility of retrospective-continuous glucose monitoring (r-CGM) in guiding insulin dosing. The new model of care features General Practitioners (GPs) and Practice Nurses (PNs) working in an expanded role, with Credentialed Diabetes Educator - Registered Nurse (CDE-RN) support. METHODS: Insulin-naïve T2DM patients (HbA1c >7.5% [>58 mmol/mol] despite maximal oral therapy) from 22 general practices in Victoria, Australia commenced insulin glargine, with glulisine added as required. Each was randomised to receive r-CGM or self-monitoring of blood glucose (SMBG). Glycaemic control (HbA1c) was benchmarked against specialist ambulatory patients referred for insulin initiation. RESULTS: Ninety-two patients mean age (range) 59 (28-77) years; 40% female; mean (SD) diabetes duration 10.5 (6.1) years participated. HbA1c decreased from (median (IQR)) 9.9 (8.8, 11.2)%; 85 (73, 99) mmol/mol to 7.3 (6.9, 7.8)%; 56 (52, 62) mmol/mol at 24 weeks (p < 0.0001). Comparing r-CGM (n = 46) with SMBG (n = 42), there were no differences in major hypoglycaemia (p=0.17) or ΔHbA1c (p = 0.31). More r-CGM than SMBG participants commenced glulisine (26/48 vs. 7/44; p < 0.001). Results were comparable to 82 benchmark patients, with similar low rates of major hypoglycaemia (2/89 vs. 0/82; p = 0.17) and less loss to follow up in the INITIATION group (3/92 vs. 14/82; p = 0.002). CONCLUSIONS: Insulin initiation and titration for T2DM patients in primary care was safe and improved HbA1c with low rates of major hypoglycaemia. CDE-RNs were effective in a new consultant role. r-CGM use in primary care was feasible and enhanced post-prandial hyperglycaemia recognition. Trial registration ACTRN12610000797077.