Objective：To stndy the correction in adopting pre-operative POSSUM and P-POSSUM for prediction of complications and mortality and  comparison with the actual postoperative complications and actual  mortality rates.
Methods：POSSUM and P-POSSUM were used for prospective study in 633 cases undergoing G1 surgery in recent 1 year. With the various indicators, and clinical effects of common factors as reference, the POSSUM predicted complication rate and P-POSSUM predicted mortality rate were compared with the actual complication and mortality rate. Logistic regression analysis，non-parametric independent samples T test (Independent-Samples T Test), area under receiver operating characteristic curve, and prediction and the actual mortality (complications) in line with the degree of (OE ratio) evaluation of statistical methods were used to assess POSSUM and P-POSSUM prediction accuracy.
Results：POSSUM predictiom in gastrointestinal surgery patients with complications was 229 cases, while the actual complications occured in 188 cases. P-POSSUM prediction of mortality was in 44 cases, while the actual deaths  occured in 34 cases;  in comparison there was no statistically significant difference (P＞0.05). In high-risk group (R1>38, there was R2>50) more prediction accuracy and the actual mortality (complications) was in line with the degree of (OE ratio) equal 1, while in low-risk group  (R1<38、R2<50) it over-estimated mortality rate, and prediction and the actual mortality (complications) in line with the degree of  evaluation was greater than 1. Logistic regression analysis of many factors, liver function, blood glucose, body height ratio, operation time and intraoperative blood pressure stable etc. related with the postoperative complications and mortality.Conclusions：For gastrointestinal surgical patients, POSSUM prediction of their complication rate and P-POSSUM prediction of mortality have a good correlation.