Aims. Our aim was to perform a meta-analysis of the outcomes of revision anterior cruciate ligament (ACL) reconstruction, comparing the use of different types of graft. Materials and Methods. A search was performed of Medline and Pubmed using the terms “Anterior Cruciate Ligament” and “ACL” combined with “revision”, “re-operation” and “failure”. Only studies that reported the outcome at a minimum follow-up of two years were included. Two authors reviewed the papers, and outcomes were subdivided into autograft and allograft. Autograft was subdivided into hamstring (HS) and bone-patellar tendon-bone (BPTB). Subjective and objective outcome measures were analysed and odds ratios with confidence intervals were calculated. Results. A total of 32 studies met the inclusion criteria. Five studies used HS autografts, eight reported using BPTB autografts, two used quadriceps tendon autografts and eight used various types. Seven studies reported using allografts, while the two remaining used both BPTB autografts and allografts. Overall, 1192 patients with a mean age of 28.7 years (22.5 to 39) and a mean follow-up of 5.4 years (2.0 to 9.6) were treated with autografts, while 269 patients with a mean age of 28.4 years (25 to 34.6) and a mean follow-up of 4.0 years (2.3 to 6.0) were treated with allografts. Regarding allografts, irradiation with 2.5 mrad was used in two studies while the graft was not irradiated in the seven remaining studies. Reconstructions following the use of autografts had better outcomes than those using allograft with respect to laxity, measured by KT-1000/2000 (MEDmetric Corporation) and the rates of complications and re-operations. Those following the use of allografts had better mean Lysholm and Tegner activity scores compared with autografts. If irradiated allografts were excluded from the analysis, outcomes no longer differed between the use of autografts and allografts. Comparing the types of autograft, all outcomes were similar except for HS grafts which had better International Knee Documentation Committee scores compared with BPTB grafts. Conclusion. Autografts had better outcomes than allografts in revision ACL reconstruction, with lower post-operative laxity and rates of complications and re-operations. However, after excluding irradiated allografts, outcomes were similar between autografts and allografts. Overall, the choice of graft at revision ACL reconstruction should be on an individual basis considering, for instance, the preferred technique of the surgeon, whether a combined reconstruction is required, the type of graft that was previously used, whether the tunnels are enlarged and the availability of allograft. Cite this article: Bone Joint J 2017;99-B:714–23