OBJECTIVE: We sought to examine potential predictors of readmission after coronary artery bypass graft (CABG) surgery. DESIGN/SETTING: We analysed routinely collected data of CABG patients who have used the public hospital system of Victoria, Australia from July 1998 to June 2003. In total, 6,627 patients were selected by linking records of elective surgery waiting time data (Elective Surgery Information System), emergency department data (Victorian Emergency Minimum Dataset) and hospital discharge data (Victorian Admitted Episodes Dataset). MEASUREMENTS: The outcome measures were 7-day, 30-day and 6-month readmissions. Possible predictors included were age, gender, Charlson comorbidity index, waiting times, length of stay in the hospital, and frequency of emergency department (ED) visits before CABG surgery. RESULTS: 7.1%, 15.2%, and 32.3% of the study population were readmitted at 7 days, 30 days and 6 months respectively. In a multivariable regression model Charlson comorbidity index was associated with 30-day (OR = 1.18; 95% CI 1.11-1.24; P < 0.01) and 6-month readmission (OR = 1.20; 95% CI 1.15-1.26; P < 0.01). Multiple ED visits were associated with 7 day (OR = 1.75; 95% CI 1.28-2.38; P < 0.01), 30 day (OR = 1.53; 95% CI 1.22-1.93; P < 0.01) and 6 month (OR = 1.80; 95% CI 1.49-2.18; P < 0.01) readmission. Waiting time was not a statistically significant predictor of readmission.