Redo coronary artery bypass grafting (CABG) forms an increasing part of the coronary revascularisation workload and currently accounts for 4% of all CABG procedures in the UK1 and 10–20% in the USA.2-4 In this issue Dougenis and Brown report their experience of redo CABG using either at least one internal mammary artery (IMA) conduit or only venous conduits.5 Their main conclusion, after a mean follow up of seven years in 103 patients, is that those who received at least one IMA graft had an improved long term outcome in terms of freedom from recurrent angina, freedom from cardiac events, and actuarial survival, compared to those receiving only vein grafts. As their study was not randomised, however, how confident can we be that important clinical differences in the groups did not contribute significantly to differences in outcome?