The main determinants of mortality after burn injury that can be measured on admission include age, total burn size (% burn), and inhalation injury (INHAL). Other variables, measured during resuscitation, may provide additional information about injury severity. We assessed the utility of early arterial blood gas (ABG) data in predicting mortality after burn injury. Data were limited to samples obtained during the first 2 days after burn injury and to those obtained during high-frequency percussive ventilation. Mean values for each patient's ABG data were calculated; subsequent analysis used these derived variables. Logistic regression analysis (LRA) was used to generate a mortality predictor using burn, age (as a cubic age score, AGE), and INHAL. LRA was then repeated with the ABG variables. A total of 162 patients were included. By univariate analysis, death was associated with increased alveolar-arterial gradient (AaDO2), AGE, % burn, full-thickness burn size, INHAL, and with decreased pH and base excess. LRA of % burn, AGE, INHAL, and full-thickness burn size retained the first three variables. The addition of ABG data demonstrated that mean burn excess and mean AaDO2 also contributed independently to mortality. However, there was no difference in accuracy (86%) between the two equations. By Kaplan Meier analysis, AaDO2 but not BE predicted earlier death in those who died. Measured during resuscitation, metabolic acidosis (ie, a base deficit) and oxygenation failure (ie, increased AaDO2) contributed independently, but modestly, to ultimate mortality after burn injury. The inclusion of these variables did not improve predictive accuracy. Whether therapies targeted at these endpoints would improve outcome is unknown.