Background This study investigated the efficacy of first-generation (cefazolin) and third-generation (ceftizoxime) prophylactic antibiotics in patients undergoing cardiac surgery and the incidence of surgical site infections, hospital stay lengths, and medical costs. Methods All adult patients (≥20 years) undergoing cardiac surgery (coronary artery bypass surgery, valve operation, or combined surgery) at one hospital from January 01, 2009 to December 31, 2016 were included in this study. A single prophylactic antibiotic was administered at a dose of 1 g within 1 hour of surgical incision and for three days after surgery at eight-hour intervals. After the propensity score matching, 194 patients in each antibiotic prophylaxis groups (first-generation vs third-generation) were analyzed. Among the 388 patients, the incidence of surgical site infection were compared according to the type of prophylactic antibiotics and risk factors were evaluated by chi-squared tests followed by multivariate logistic regression analysis. A Student’s t-tests were analyzed to compare hospitalization and medical costs. Results The incidence of deep surgical site infections significantly lower in first-generation group (5.7%) than third-generation group (16.5%). The pathogens isolated from surgical infection sites were similarly distributed in both groups, but gram-positive bacteria were more highly infectious than gram-negative bacteria (67% vs 23%). Preoperative hospitalization duration, mean operation time, and ventilator use time were similar in both groups but the postoperative hospitalization duration was significantly shorter in the first-generation group (25.5 days) than third-generation (29.8 days). In addition, the medical cost lower in the first-generation group (20,594 USD) than third-generation (26,488 USD). Conclusion In conclusion, the first-generation (cefazolin) is better than the third-generation (ceftizoxime) as a prophylactic antibiotic in reducing surgical site infection rates, hospitalization lengths, and medical expenditures.