Aim Recent studies have revealed positive effects of endometrial injury on clinical pregnancy rates, but with inconsistent results. The aim of this meta-analysis was to assess the efficacy of endometrial injury (biopsy and/or hysteroscopy) as a potential treatment measure for implantation failure in the in vitro fertilization population. Methods We searched in PubMed for studies comparing the efficacy of endometrial injury versus no intervention in women with at least one implantation failure. A random-effects model was used to evaluate the clinical pregnancy rate (CPR) and the live birth rate (LBR). Results Seventeen studies (11 randomized and 6 non-randomized studies) were included in this meta-analysis, including 1864 women in the intervention group and 2193 women in the control group. Overall, the CPR and LBR were significantly higher in the intervention group than in the control group (for CPR, n = 3997, 16 studies, P < 0.00001, risk ratio (RR) = 1.47, 95% confidence interval [CI] [1.24, 1.74]; for LBR, n = 2361, 11 studies, P = 0.003, RR = 1.41, 95% CI [1.12, 1.78]). However, after excluding the non-randomized studies, there was no significant difference in CPR (P = 0.29 for endometrial biopsy, P = 0.05 for hysteroscopy) and LBR (P = 0.23 for endometrial biopsy, P = 0.39 for hysteroscopy) between the intervention group and the control group. Conclusion There is still insufficient evidence to support the use of endometrial injury in women with implantation failure. Robust randomized controlled studies should be designed and performed before clinical implementation of endometrial injury.