We critically evaluated several diagnostic modalities (clinical criteria, subxiphoid pericardial window (SPW) and laparoscopy) used in the detection of occult cardiac injury in haemodynamically stable patients at high risk of cardiac injury. Over 5 years, 122 patients were admitted to a Level I trauma centre with such an injury. They sustained 69 stab wounds, and 53 gunshot wounds. Sites of penetration were: precordial (81),right chest (25), lateral chest (13), thoracoabdominal (40) and abdominal (19). Vital signs in the emergency room were (mean +/- SD): systolic BP, 111 +/- 23.2 mmHg; HR, 106 +/- 18.7; GCS, 13.6 +/- 1.3; and CVP, 17 +/- 7.8 cmH(2)O. SPW was performed in ail patients and was positive for haemopericardium in 26 patients, 24 (92 per cent) of whom had a cardiac injury at operation. Two patients had pericardial lacerations without cardiac injury. In addition, 14 patients with lower precordial and thoracoabdominal wounds underwent laparoscopy. At laparoscopy, the pericardium teas evaluated by transdiaphragmatic inspection in 10 patients. The presence (two) or absence (eight) of blood within the pericardium runs accurately predicted and verified by SPW. Univariate and multiple logistic regression analysis of clinical data failed to reveal any significant predictor of cardiac injury. SPW remains the standard means of diagnosing occult cardiac injury in high-risk patients. Since the incidence of occult cardiac injury in haemodynamically stable patients is 20 per cent, SPW should be used liberally. Laparoscopy may have a role in evaluating the pericardium in the subgroup of patients with lower chest wounds, and if facilitates inspection of intra-abdominal viscera and diaphragm at the same time.