Measurement of gastric intramucosal pH (pHi) has been advocated to assess gastric perfusion. Regional PCO2 (rPCO2) values are measured using saline tonometry (rsPCO2) and more recently using air tonometry (raPCO2). We compared 237 measurements of saline and air tonometry in 19 consecutive, severely ill patients (mean age 59 (range 31-76) yr, 19 males, APACHE II 22 +/- 7) with cardiogenic shock. Equilibration period was set to 90 min. Nineteen independent paired samples of mean raPCO2 and mean rsPCO2 of each patient showed good correlation (r = 0.93, P < 0.001). Mean raPCO2 was 6.5 (1.8) kPa and mean rsPCO2 6.8 (2.4) kPa. PCO2 measured by saline was significantly higher than that measured by air (P < 0.05). Bland and Altman analysis showed a bias (mean rsPCO2-mean raPCO2) of 0.3 kPa and a precision of 1.2 kPa. Agreement between the two methods decreased with increasing rPCO2 concentrations. Although air tonometry of rPCO2 is a promising technique, a systematic disagreement with saline tonometry at high rPCO2 values requires further investigation and cautious interpretation of these values.