To determine the ability of prehospital end-tidal carbon dioxide (ETCO 2 ) to predict in-hospital mortality compared to conventional vital signs.We conducted a retrospective cohort study among patients transported by emergency medical services during a 29-month period. Included patients had ETCO 2 recorded in addition to initial vital signs. The main outcome was death at any point during hospitalization. Secondary outcomes included laboratory results and admitting diagnosis.Of 1328 records reviewed, hospital discharge data, ETCO 2 , and all 6 prehospital vital signs were available in 1088 patients. Low ETCO 2 levels were the strongest predictor of mortality in the overall group (area under the receiver operating characteristic curve (AUC of 0.76, 95% confidence interval [CI] 0.66-0.85), as well as subgroup analysis excluding prehospital cardiac arrest (AUC of 0.77, 95% CI 0.67-0.87). The sensitivity of abnormal ETCO 2 for predicting mortality was 93% (95% CI 79%-98%), the specificity was 44% (95% CI 41%-48%), and the negative predictive value was 99% (95% CI 92%-100%). There were significant associations between ETCO 2 and serum bicarbonate levels (r = 0.429, P < .001), anion gap (r = −0.216, P < .001), and lactate (r = –0.376, P < .001).Of all prehospital vital signs, ETCO 2 was the most predictive and consistent for mortality, which may be related to an association with metabolic acidosis.