Background : Risk factors for surgical site infection (SSI) are thought to include poorly controlled diabetes mellitus, dialysis, and a long operating time, but patients without risk factors can also develop infection. Therefore, it is possible that SSI could be prevented by routinely using the precautions against SSI developed for high-risk patients. We investigated the route and pathogenetic mechanism of mediastinitis, which is the most frequent SSI after cardiac surgery. We hypothesized that mediastinitis occurred when the deep sternal marrow was contaminated by skin bacteria. Based on this hypothesis, we investigated the efficacy of various intraoperative prophylactic methods for preventing mediastinitis. Methods : We evaluated 658 patients undergoing cardiac surgery at our institution between April 2011 and July 2016. They were classified into two groups. Group C comprised 406 patients who received standard insertion of a sternal retractor after sternotomy. Group S was 252 patients in whom a retractor was inserted after covering the sternal marrow with oxidized cellulose hemostats and belt-like thin towels, with surplus parts of the towels being used to fill subcutaneous dead space at the superior and inferior margins of the midline wound. We investigated the following 10 risk factors for mediastinitis: diabetes (HbA1c≥7.5), renal failure (Cr≥2), smoking, obesity (BMI≥30), reoperation, urgent/emergency operation, intubation in the preoperative period, long operating time (≥8 h), reopening the chest for hemostasis, and coronary artery bypass grafting (CABG). Factors associated with mediastinitis were determined using univariate modeling analysis followed by multi-variate logistic regression analysis. Results : Mediastinitis occurred in 13 patients (2.0%). The significant risk factor for mediastinitis were urgent/emergency operation and CABG, but 1 patient had no risk factors. A univariate analysis showed statistical significance in CABG, presence of maneuver covering the sternal marrow, JapanSCORE-II in mortality and deep sternum infection (DSI). Reopening the chest for hemostasis, CABG, aortic aneurysm, plural risk factors, and JapanSCORE-II in DSI were identified as a risk factor by multiple logistic regression, not all factors showed a significant difference. Mediastinitis only occurred in group C, and it was significantly less frequent in group S with additional precautions against infection including propensity score matching analysis (p<0.05). Conclusion : When the bone marrow of the transected sternum was covered tightly to protect it from contamination by skin bacteria during cardiac surgery, the frequency of postoperative mediastinitis was significantly reduced.