Objective To identify the predictors for readmission in the ICU among cardiac surgery patients. Methods We conducted a retrospective cohort study of 2 799 consecutive patients under cardiac surgery, who were divided into two groups including a readmission group (47 patients, 27 males and 20 females at age of 62.0±14.4 years) and a non readmission group (2 752 patients, 1 478 males and 1 274 females at age of 55.0±13.9 years) in our hospital between January 2014 and October 2016. Results The incidence of ICU readmission was 1.68% (47/2 799). Respiratory disorders were the main reason for readmission (38.3%).Readmitted patients had a significantly higher in-hospital mortality compared to those requiring no readmission (23.4% vs. 4.6%, P<0.001). Logistic regression analysis revealed that pre-operative renal dysfunction (OR=5.243, 95%CI 1.190 to 23.093, P=0.029), the length of stay in the ICU (OR=1.002, 95%CI 1.001 to 1.004, P=0.049), B-type natriuretic peptide (BNP) in the first postoperative day (OR=1.000, 95%CI 1.000 to 1.001, P=0.038), acute physiology and chronic health evaluationⅡ (APACHEⅡ) score in the first 24 hours of admission to the ICU (OR=1.171, 95%CI 1.088 to1.259, P<0.001), and the drainage on the day of surgery (OR=1.001, 95%CI1.001 to 1.002, P<0.001) were the independent risk factors for readmission to the cardiac surgery ICU. Conclusion The early identification of high risk patients for readmission in the cardiac surgery ICU could encourage both more efficient healthcare planning and resources allocation.