Purpose: Pulse oximetry is widely used to monitor respiratory status during ERCP, but does not detect hypoventilation (CO2 retention), especially in patients receiving supplemental oxygen. The purpose of this study was to determine whether transcutaneous CO2 (PtcCO2) monitoring during ERCP could prevent severe CO2 retention or other adverse events. Methods: All patients received automated hemodynamic/cardiac monitoring, pulse oximetry, and PtcCO2 monitoring. Level of sedation and respiratory rate were closely assessed and recorded every five minutes. Patients were randomized into a treatment arm (Group 1) where PtcCO2 data was available to guide sedation during the procedure or a blinded arm where it was not (Group 2). Sedation was withheld in Group 1 for PtcCO2 greater than 25 mmHg over baseline, or actively rising. Oxygen was administered for oxygen saturation < 90%. Multivariate analysis was used to determine independent predictors of peak and maximum change in PtcCO2. Results: 395 patients were randomized. There were no significant differences between the groups at baseline with respect to patient demographics, comorbid conditions, or monitoring parameters. Overall, 10% of patients reached a PtcCO2 of 60 mmHg or greater, including 3 patients reaching 100, 102, and 105 mmHg. Mean rise in PtcCO2 level in Group 1 was 10.8+0.6 mmHg, and in Group 2 was 11.7+0.6 mmHg (NS). No patient in Group 1 experienced a rise in CO2 greater than 40 mmHg above baseline, compared with 5/196 (2.5%) patients in the blinded group (p=0.03). Baseline PtcCO2 level was the only independently significant pre-randomization predictor of peak PtcCO2; age, comorbid conditions, alcohol or tobacco use, or baseline oxygen were not. Post-randomization variables associated with peak CO2 level included combined benzodiazepine/narcotic sedation, degree of oxygen desaturation, amount of oxygen administered, and use of naloxone. Correlations between clinical monitoring and objective measures of respiratian were poor: deepest sedation vs. PtcCO2 (R=0.3) or SpO2 (R=0.06); slowest respiratory rate vs. PtcCO2 (R=0.4) or SpO2 (R=0.5). Conclusions: CO2 retention during ERCP is common, may be severe, and is poorly detected by close clinical assessment. Although there were no cardioresp. arrests in this series, the use of transcutaneous CO2 monitors to guide sedation did prevent severe CO2 retention. The cost-effectiveness of this technology remains to be determined.