Objective To identify the risk factors associated with prolonged length of stay (LOS) despite an enhanced recovery after surgery (ERAS) protocol in short-level lumbar fusion surgery. Methods We gathered data for all patients undergoing short-level lumbar fusion surgery from January to November 2021. Given the discharge criteria, a threshold was set according to mean LOS, and two groups were spontaneously formed: LOS shorter than the threshold for discharge (control group, n = 114) and LOS longer or equal to the threshold for discharge (delayed group, n = 72). Preoperative metrics were compared to identify risk factors associated with prolonged LOS. Results A total consecutive 186 patients with complete medical records were enrolled (77 males and 109 females; mean age 71.08 ± 5.70 years). After dichotomization according to the threshold for discharge, there were 114 patients in control group and 72 in delayed group. Statistical analysis demonstrated that age ≥75 years ( P = .002), female sex ( P < .001), American Society of Anesthesiologists grade ≥ 3 ( P = .035), operation time ( P < .001), anesthesia time ( P < .001), ambulation time >1 day ( P = .027), removal of urinary catheter time >1 day ( P = .019), fusion levels ( P < .001), Clavien-Dindo grade > 1 ( P <.001) and allogeneic transfusion ( P = .009) were significantly related to prolonged LOS. Binary logistic regression revealed that age (odds ratio (OR) 5.149; 95% confidence interval (CI) 2.045-12.966, P = .001), sex (OR 5.185, 95% CI 2.183-12.317, P < .001) and Clavien-Dindo grade > 1(OR 15.936, 95% CI 5.220-48.652, P < .001) were independent risk factors of prolonged LOS. Conclusions In this retrospective study, we analyzed the potential risk factors associated with delayed LOS despite implementing ERAS, further, binary logistic regression exhibited that age ≥75 years old, female sex and Clavien-Dindo grade >1 were independently correlated with prolonged LOS.