OBJECTIVE To determine the effect of graft choice (allograft, bone-patellar tendon-bone autograft, or hamstring autograft) on deep tissue infections following anterior cruciate ligament (ACL) reconstructions. DESIGN Retrospective cohort study. SETTING AND POPULATION Patients from 6 US health plans who underwent ACL reconstruction from January 1, 2000, through December 31, 2008. METHODS We identified ACL reconstructions and potential postoperative infections using claims data. A hierarchical stratified sampling strategy was used to identify patients for medical record review to confirm ACL reconstructions and to determine allograft vs autograft tissue implanted, clinical characteristics, and infection status. We estimated infection rates overall and by graft type. We used logistic regression to assess the association between infections and patients' demographic characteristics, comorbidities, and choice of graft. RESULTS On review of 1,452 medical records, we found 55 deep wound infections. With correction for sampling weights, infection rates varied by graft type: 0.5% (95% CI, 0.3%-0.8%) with allografts, 0.6% (0.1%-1.5%) with bone-patellar tendon-bone autografts, and 2.5% (1.9%-3.1%) with hamstring autograft. After adjusting for potential confounders, we found an increased infection risk with hamstring autografts compared with allografts (odds ratio, 5.9; 95% CI, 2.8-12.8). However, there was no difference in infection risk among bone-patellar tendon-bone autografts vs allografts (odds ratio, 1.2; 95% CI, 0.3-4.8). CONCLUSIONS The overall risk for deep wound infections following ACL reconstruction is low but it does vary by graft type. Infection risk was highest in hamstring autograft recipients compared with allograft recipients and bone-patellar tendon-bone autograft recipients.