Background Surgical wound classification (SWC) is used by hospitals, quality collaboratives, and Centers for Medicare and Medicaid to stratify patients for their risk for surgical site infection. Although these data can be used to compare centers, the validity and reliability of SWC as currently practiced has not been well studied. Our objective was to assess the reliability of SWC in a multicenter fashion. We hypothesized that the concordance rates between SWC in the electronic medical record and SWC determined from the operative note review is low and varies by institution and operation. Study Design Surgical wound classification concordance was assessed at 11 participating institutions between SWC from the electronic medical record and SWC from operative note review for 8 common pediatric surgical operations. Cases with concurrent procedures were excluded. A maximum of 25 consecutive cases were selected per operation from each institution. A designated surgeon reviewed the included operative notes from his/her own institution to determine SWC based on a predetermined algorithm. Results In all, 2,034 cases were reviewed. Overall SWC concordance was 56%, ranging from 47% to 66% across institutions. Inguinal hernia repair had the highest overall median concordance (92%) and appendectomy had the lowest (12%). Electronic medical records and reviewer SWC differed by up to 3 classes for certain cases. Conclusions Surgical site infection risk stratification by SWC, as currently practiced, is an unreliable methodology to compare patients and institutions. Surgical wound classification should not be used for quality benchmarking. If SWC continues to be used, individual institutions should evaluate their process of assigning SWC to ensure its accuracy and reliability.