Abstract Aim We tested the hypothesis that metformin may regress left ventricular hypertrophy (LVH) in patients who have coronary artery disease (CAD), with insulin resistance (IR) and/or pre-diabetes. Methods and results We randomly assigned 68 patients (mean age 65 ± 8 years) without diabetes who have CAD with IR and/or pre-diabetes to receive either metformin XL (2000 mg daily dose) or placebo for 12 months. Primary endpoint was change in left ventricular mass indexed to height1.7 (LVMI), assessed by magnetic resonance imaging. In the modified intention-to-treat analysis (n = 63), metformin treatment significantly reduced LVMI compared with placebo group (absolute mean difference −1.37 (95% confidence interval: −2.63 to −0.12, P = 0.033). Metformin also significantly reduced other secondary study endpoints such as: LVM (P = 0.032), body weight (P = 0.001), subcutaneous adipose tissue (P = 0.024), office systolic blood pressure (BP, P = 0.022) and concentration of thiobarbituric acid reactive substances, a biomarker for oxidative stress (P = 0.04). The glycated haemoglobin A1C concentration and fasting IR index did not differ between study groups at the end of the study. Conclusion Metformin treatment significantly reduced LVMI, LVM, office systolic BP, body weight, and oxidative stress. Although LVH is a good surrogate marker of cardiovascular (CV) outcome, conclusive evidence for the cardio-protective role of metformin is required from large CV outcomes trials.