Background: The choice of anesthesia technique remains debatable in patients undergoing surgical repair of hip fracture. This meta-analysis was performed to compare the effect of neuraxial (epidural/spinal) versus general anesthesia on perioperative outcomes in patients undergoing hip fracture surgery. Methods: Medline, Cochrane Library, Science-Direct, and EMBASE databases were searched to identify eligible studies focused on the comparison between neuraxial and general anesthesia in hip fracture patients between January 2000 and May 2019. Perioperative outcomes were extracted for systemic analysis. Sensitivity analyses were conducted using a Bonferroni correction and the leave-one-out method. The evidence quality for each outcome was evaluated by the Grading of Recommendations Assessment, Development and Evaluation (GRADE) system. Results: Nine randomized controlled trials (RCTs) including 1084 patients fulfilled our selection criteria. The outcomes for the meta-analysis showed that there were no significant differences in the 30-day mortality (OR = 1.34, 95% CI 0.56, 3.21; P = 0.51), length of stay (MD = − 0.65, 95% CI-0.32, 0.02; P = 0.06), and the prevalence of delirium (OR = 1.05, 95% CI 0.27, 4.00; P = 0.95), acute myocardial infarction (OR = 0.88, 95% CI 0.17, 4.65; P = 0.88), deep venous thrombosis (OR = 0.48, 95% CI 0.09, 2.72; P = 0.41), and pneumonia (OR = 1.04, 95% CI 0.23, 4.61; P = 0.96) for neuraxial anesthesia compared to general anesthesia, and there was a significant difference in blood loss between the two groups (MD = − 137.8, 95% CI-241.49, − 34.12; p = 0.009). However, after applying the Bonferroni correction for multiple testing, all the adjusted p-values were above the significant threshold of 0.05. The evidence quality for each outcome evaluated by the GRADE system was low. Conclusions: In summary, our present study demonstrated that there might be a difference in blood loss between patients receiving neuraxial and general anaesthesia, however, this analysis was not robust to adjustment for multiple testing and therefore at high risk for a type I error. Due to small sample size and enormous inconsistency in the choice of outcome measures, more high-quality studies with large sample size are needed to clarify this issue.