Objectives To evaluate the effects of hepatitis B immunoglobulin (HBIG) intrauterine injection before delivery on interrupting mother-to-child transmission of hepatitis B virus (HBV). Methods The randomized controlled trials (RCTs) on HBIG intrauterine injection on interrupting mother-to-child transmission of HBV published between January 1992 and May 2012 were searched in The Cochrane Library, PubMed, CBM, CNKI, VIP, WanFang Data, etc. The studies were screened according to inclusive and exclusive criteria, the data were extracted, the quality was assessed by two reviewers independently, and meta-analysis, publication bias and sensitivity analysis was conducted using Stata software. Results The total 42 studies involving 7 212 infants were included. The randomized methods were asserted in all studies, three of which reported the details of randomization, one study mentioned blinded method, two studies mentioned incomplete outcome data, 13 studies had other potential threats to validity, no allocation concealment and selective outcome reporting was mentioned. Results of meta-analysis indicated that the infant HBV infection rates in the HBIG group and the control group were 8.971% and 25.470% (RR=0.359, 95%CI 0.303 to 0.425) at birth, 5.385% and 13.919% (RR=0.391, 95%CI 0.278 to 0.550) after half a year, 5.318% and 12.457% (RR=0.429, 95%CI 0.335 to 0.551) after one year; the infant anti-HBs rates in the HBIG group and the control group were 61.964% and 14.523% (RR=6.712, 95%CI 1.920 to 23.467) at birth, 77.754% and 66.311% (RR=1.209, 95%CI 0.989 to 1.478) after half a year. Funnel graphs showed that there was publication bias. Sensitivity analysis showed that the results except the infant anti-HBs protection after half-a-year follow-up were stable and consistent with the original results. Conclusion Injection of HBIG during pregnancy for HBV-carrying mothers can effectively reduce the occurrence of HBV whenever at birth, after half a year or after one year, and increase the infant anti-HBs protection rate at birth, but it is ineffective to improve anti-HBV protection rate after half a year. Owing to the low quality of the included studies and existence of biases, this conclusion should be cautiously put into clinical practice.