To construct a predictive model for a prolonged stay in the intensive care unit (ICU) for coronary artery bypass graft surgery (CABG). Eight hundred and eighty-eight patients undergoing CABG were studied by univariate and multivariate analysis. Prolonged stay in the ICU was defined as >/=3 days stay. Stepwise selective procedure (P/=0.40 was used as cut-off point for the prognostic test. The specificity of this test for prolonged stay in the ICU was 99%; sensitivity 9%; positive predictive value 60%; and negative predictive value 89%. The results show that individual patients presented for CABG, can be stratified according to their risk for prolonged stay >/=3 days in the ICU.