Objective To screen the risk factors of postoperative cognitive dysfunction (POCD) in elderly patients undergoing gastrointestinal tumor resection.Methods A total of 221 elderly patients scheduled for elective gastrointestinal tumor resection under general anesthesia from June 2018 to June 2020 were selected. Cognitive function of each participant was assessed by Mini-mental State Examination (MMSE). According to whether POCD occurred, patients were divided into non-POCD group and POCD group, while the latter was further subdivided into mild POCD group and severe POCD group on the basis of the severity of disease conditions. Clinical data of three groups of patients before, during and after the operation were recorded separately, and the risk factors for POCD were analyzed by univariate and multivariate Logistic regression models.Results Of the 221 patients, 91 (41.2%) had POCD, and 40 (18.1%) were classified as severe POCD. The frequency of history of smoking and alcohol consumption was higher in POCD group than that in non-POCD group (P < 0.05). There was no difference in age, gender composition, body mass index (BMI), American Society of Anesthesiologists (ASA) physical status, education degree, preoperative MMSE score, and underlying medical history between the POCD group and non-POCD group (P >0.05). Preoperative K+ level and EuroQol five dimensions questionnaire (EQ-5D) score 30 days after surgery were lower in POCD group than those in non-POCD group (P < 0.05). The incidences of pain at 1 day after surgery [Visual Analogue Scale (VAS) scores at rest or on movement ≥ 4], intraoperative blood loss ≥1,000 ml, and postoperative requirement for treatments in intensive care unit (ICU) were higher in POCD group compared with non-POCD group (P < 0.05). No difference was observed in preoperative grip strength, white blood cell (WBC) count, hemoglobin and albumin levels, operative duration, the choice of transversus abdominis plane block, intraoperative warming or use of nonsteroidal anti-inflammatory drugs and dexmedetomidine between the two groups (P >0.05). The rates of hypertension, postoperative requirement for treatments in ICU, preoperative WBC count ≥10×10⁹/L and intraoperative blood loss ≥500 ml were different among severe POCD group, mild POCD group, and non-POCD group (P > 0.05). Univariate Logistic regression analysis indicated that VAS scores ≥4 at rest and on movement at 1 day after operation, history of alcohol consumption and intraoperative blood loss ≥1,000 ml were risk factors for early POCD (P < 0.05). Besides, VAS scores at rest and on movement ≥ 4 at 1 day after operation, preoperative WBC count ≥10×10⁹/L and intraoperative blood loss ≥ 500 ml were risk factors for severe POCD (P <0.05). Multivariate Logistic regression analysis showed that VAS score at rest ≥ 4 at 1 day after operation [R = 6.445, (95% CI: 2.621, 15.845)], VAS score on movement ≥ 4 at 1 day after operation [R = 6.037, (95% CI: 2.996, 12.164)], history of alcohol consumption [R = 2.320, (95% CI: 1.091, 4.937)] and intraoperative blood loss ≥1000 ml [R = 11.631, (95% CI: 1.172, 115.464)] were risk factors for early POCD, while VAS score at rest ≥4 [R = 9.583, (95% CI: 3.086, 29.762)] at 1 day after operation, VAS score on movement ≥4 [R = 7.655, (95% CI：2.576, 22.743)] at 1 day after operation, preoperative WBC count ≥10×10⁹/L [R = 11.952, (95% CI: 1.686, 84.758)], intraoperative blood loss ≥ 500 ml [R = 7.235, (95% CI: 1.943, 26.942)] and history of hypertension [R = 2.753, (95% CI: 1.103, 6.866)] were risk factors for severe POCD.Conclusions High VAS score at 1 day after operation, history of alcohol consumption and intraoperative blood loss more than 1,000 ml are risk factors for early POCD in elderly patients undergoing gastrointestinal tumor resection. In addition, high VAS scores at 1 day after operation, preoperative WBC count exceeds 10×10⁹/L, intraoperative blood loss greater than 500 ml and history of hypertension are risk factors for severe POCD in elderly patients undergoing gastrointestinal tumor resection.