One hundred five hemodynamically stable patients with penetrating thoracic trauma were prospectively evaluated for occult cardiac injury. All patients underwent transthoracic echocardiography (ECHO) and subxiphoid exploration (SXE). Those with positive SXE underwent exploration. Results are calculated based on operative findings of significant injuries. For the entire group, SXE was more sensitive than ECHO (sensitivity, 100% vs. 56%; specificity, 92% vs. 93%; and accuracy, 92% vs. 90%, respectively). The SXE missed no injuries in this group. Eight of 17 patients, however, with positive SXEs had insignificant injuries at sternotomy. Seven patients had positive ECHO, but no significant injury. The ECHO missed four significant injuries. When comparing SXE with ECHO in patients without hemothorax, however, sensitivity (100% vs. 100%), specificity (89% vs. 91%), and accuracy (90% vs. 91%) were comparable between SXE and ECHO. We conclude that ECHO has significant limitations in identifying serious cardiac injuries in patients with hemothorax. For hemodynamically stable patients without hemothorax, ECHO missed no significant injuries and may be an acceptable diagnostic option for detecting significant cardiac trauma in patients with injuries in proximity to the heart.