Hemorrhagic shock (HS) is a setting in which both pulmonary and cutaneous perfusion may be impaired. The goals of this study were to evaluate the relationship between end-tidal (etCO2), transcutaneous (tPCO2), arterial carbon dioxide (PaCO2) and lactate during lethal HS and to assess the effect of progressive HS on those variables and on a new variable, the noninvasive CO2 gradient ([NICO2G] or the difference between tPCO2 and etCO2). Ten consciously sedated swine were hemorrhaged, by means of a computerized exponential protocol, of up to 80% estimated blood volume for 20 min. End-tidal carbon dioxide, tPCO2, PaCO2, and lactate measurements were taken at baseline and every 5 min thereafter, that is, after 25%, 44%, and 62% total blood volume hemorrhage (TBVH) and at cardiac arrest. Cardiac arrest occurred on average at 67% TBVH. Data were analyzed by linear regression and one-way repeated-measures analysis of variance and are presented as means ± SD. Forty-nine paired measurements were made. There was no overall relationship between NICO2 variables and PaCO2: PaCO2 vs. tPCO2 (r2 = 0.002, P = 0.78); PaCO2 vs. etCO2 (r2 = 0.0002, P = 0.93). Rather, NICO2G increased at each level of blood loss: 4.0 ± 24.9 at baseline, 6.3 ± 35.7 at 25% TBVH, 25.0 ± 37.6 at 44% TBVH, 55.0 ± 33.9 at 62% TBVH, and 70.0 ± 33.2 at cardiac arrest (P < 0.05). Similarly, tPCO2 increased and etCO2 decreased at each level. Linear regression of NICO2G and lactate showed a better correlation than was observed for the other two variables: NICO2G, r2 = 0.58; tPCO2, r2 = 0.46; etCO2, r2 = 0.26. During HS, NICO2 monitors lose accuracy for approximating the PaCO2 but gain usefulness as hemodynamic monitors. Also, by combining data from two different organ systems, NICO2G demonstrated improved correlation with lactate than did either etCO2 or tPCO2 alone.