Methods CHAOS was a prospective, double-blind, placebo-controlled, randomised, single-centre trial in the East Anglian region of the UK. Patients were recruited at one centre (Papworth Hospital) The study compared two parallel groups of patients with angiographically proven coronary atherosclerosis. One group received a-tocopherol and the other placebo. The primary outcome variables were a combined endpoint of cardiovascular death and non-fatal MI, and non-fatal MI alone. Enrolment began on Oct 10, 1992, and ended on Dec 15, 1994. The analysis included all endpoints between Oct 10, 1992, and June 18, 1995. The CHAOS trial was designed to have 80% power (with 2p=0-05) to detect a relative risk of the combined endpoint (nonfatal MI and cardiovascular death) of less than 0-75 between the treatment groups after median follow-up of 1-5 years. This calculation assumed an accrual rate of 1000 individuals per year for 2 years, and an event rate of 5% per year, which were estimated from past referral and event rates at Papworth Hospital. Blinded interim analysis was planned for safety reasons, with trial termination if the relative risk between the treatment groups was below 0-70 with 2p