We evaluated the effects of CO2 embolism on end-tidal carbon dioxide (ETCO2) and compared four methods for detection of gas embolism. Fourteen pigs were monitored for CO2 embolism with transesophageal echocardiography (TEE), changes in ETCO2, changes in mean pulmonary artery pressure (MPAP), and precordial auscultation (AUSC). Serial injections of CO2 (ranging from 0.05 to 5.0 mL/kg) were performed in seven pigs (Group 1). In the other seven pigs, CO2 was infused at rates between 0.01 and 0.4 mL.kg-1.min-1 (Group 2). Positive responses were defined as an acute change in heart sounds (AUSC), visualization of gas bubbles in the right cardiac chambers on TEE, an increase in MPAP > or = 3 mm Hg, and an acute change (increase or decrease) in ETCO2 > or = 3 mm Hg. In both groups, positive responses to CO2 embolism were represented by an initial decrease in ETCO2. The frequency with which positive responses were observed revealed that TEE was the most sensitive method (P < 0.05), whereas no differences were found among the other methods. In conclusion, in this model, positive response to CO2 embolism was represented by a decrease in ETCO2. TEE was the most sensitive method of detection of CO2 embolism, and ETCO2, MPAP, and AUSC were equally sensitive.