Background: The enhanced recovery after surgery (ERAS) protocol is an evidence-based perioperative care program aimed at reducing surgical stress response and accelerating recovery. However, a small propor- tion of patients fail to benefit from the ERAS program following pancreaticoduodenectomy. This study aimed to identify the risk factors associated with failure of ERAS program in pancreaticoduodenectomy. Methods: Between May 2014 and December 2017, 176 patients were managed with ERAS program fol- lowing pancreaticoduodenectomy. ERAS failure was indicated by prolonged hospital stay, unplanned read- mission or unplanned reoperation. Demographics, postoperative recovery and compliance were compared of those ERAS failure groups to the ERAS success group.Results: ERAS failure occurred in 59 patients, 33 of whom had prolonged hospital stay, 18 were readmit- ted to hospital within 30 days after discharge, and 8 accepted reoperation. Preoperative American Society of Anesthesiologists (ASA) score of ≥III (OR = 2.736; 95％ CI: 1.276-6.939; P = 0.028) and albumin (ALB) level of < 35 g/L (OR = 3.589; 95％ CI: 1.403-9.181; P = 0.008) were independent risk factors associated with prolonged hospital stay. Elderly patients ( > 70 years) were on a high risk of unplanned reoperation (62.5％ vs. 23.1％, P = 0.026). Patients with prolonged hospital stay and unplanned reoperation had delayed intake and increased intolerance of oral foods. Prolonged stay patients got off bed later than ERAS success patients did (65 h vs. 46 h, P = 0.012). Unplanned reoperation patients tended to experience severer pain than ERAS success patients did (3 score vs. 2 score, P = 0.035). Conclusions: Patients with high ASA score, low ALB level or age > 70 years were at high risk of ERAS failure in pancreaticoduodenectomy. These preoperative demographic and clinical characteristics are im- portant determinants to obtain successful postoperative recovery in ERAS program.