Sixty-five patients who arrived in the emergency department in cardiac arrest were studied prospectively to determine whether central venous pH could be used as an accurate predictor of arterial pH in prolonged cardiac arrest. Central venous and arterial access were obtained as soon as possible after arrival in the emergency department. Simultaneous arterial and venous samples were drawn and sent for blood gas analysis. Under normal conditions, central venous pH (pHcv) approximates arterial pH (pHa). In prolonged cardiac arrest, however, our data reveal a mean pHa of 7.31 +_ 0.25 and a mean pHcv of 7.08 +_ 0.19. There was moderate correlation between pHa and pHcv when all patients were considered (r = .69, P < .01). The correlation was excellent, however, in the subgroup of 15 patients who had a pulse at some point during resuscitation (r = .95, P < .01). In 13 of these 15 patients the acid base status would have been managed correctly based on the predicted pHa (pHcv + 0.12 correction factor). The pHcv was also valuable in identifying a second subgroup of patients who required no further bicarbonate therapy; all patients who had a pHcv >t 7.15 had a pHa > 7.30 (21 patients). The central venous pH was found to be a useful index of arterial pH when applied to a definable subset of patients, which in this study constituted 45% of all patients in prolonged cardiac arrest.