Cardiac arrest during spinal anesthesia is a rare event, but when it does happen cardiopulmonary resuscitation (CPR) is often ineffectual. This study examines the effect of spinal anesthesia on coronary perfusion pressure (CPP) during CPR and the subsequent response of CPP to epinephrine administration. Twenty mongrel dogs were anesthetized, and randomly assigned to a spinal injection with either 0.5 mg/kg bupivacaine or with an equivalent volume of normal saline. Twenty minutes later, ventricular fibrillation was electrically induced and after 1 min CPR was started. CPP was measured every minute. After 4 min of CPR, epinephrine 0.01 mg/kg was given followed by 0.1, 0.2, and 0.4 mg/kg epinephrine intravenously (IV) at 6, 8, 10 min of CPR, respectively. The bupivacaine (n = 11) group had significantly less CPP than the sham spinal (n = 8) group, 12-13 mm Hg as compared to 27-34 mm Hg. Only 4/11 dogs (36%) in the bupivacaine group had CPP > or = 15 mm Hg during the first 4 min after arrest as compared to 8/8 (100%) in the sham spinal group. This increased to 7/11 dogs (64%) after 0.01 mg/kg epinephrine and to 9/11 after 0.1 mg/kg epinephrine. Total spinal anesthesia decreases CPP and thus the efficacy of CPR in dogs below the threshold previously established for predicting successful resuscitation. Epinephrine is effective in increasing CPP during CPR above the critical threshold. These data suggest that if cardiac arrest occurs during spinal anesthesia, epinephrine should be given in doses of 0.01-0.02 mg/kg IV initially and then increasing to 0.1 mg/kg IV. When this does not work, and ineffective CPR is suspected, alternative resuscitative measures should be considered.