Mixed venous oxygen saturation (MVO2) is a valuable parameter in monitoring critically ill patients because it serves as an index of the adequacy of the oxygen delivery system. Mixed venous oxygen saturation as reflected by the easily obtainable central venous oxygen saturation (CVO2) may prove useful during cardiopulmonary resuscitation (CPR) as an indicator of both the adequacy of varying CPR regimens and the efficacy of pharmacological interventions. This study investigates the relationship between CVO2 and MVO2 and its clinical usefulness during CPR. Swan-Ganz and central venous catheters were placed in 20 mongrel dogs. Ventricular fibrillation was then induced and, after predetermined downtimes ranging from 5 to 60 minutes, thoracotomy was performed, and open-chest bimanual cardiac massage was started. Central venous and mixed venous blood-gas samples were drawn every five minutes during a 30-minute period of CPR. The correlation between CVO2 and MVO2 was 0.8719 (P less than 0.001) before arrest but deteriorated at all times during CPR with values ranging from 0.1589 (P = 0.542) to 0.5781 (P = 0.024). Although statistically significant at times, the correlation between CVO2 and MVO2 during CPR is not consistently high enough to enable the routine substitution of CVO2 for MVO2 in assessing the oxygen delivery system.