Target values for arterial carbon dioxide tension (PaCO2) in extracorporeal membrane oxygenation (ECMO) for acute respiratory distress syndrome (ARDS) are unknown. We hypothesized that lower PaCO2 values on ECMO would be associated with lighter sedation. We used data from two independent patient cohorts with ARDS spending 1,177 days (discovery cohort, 69 patients) and 516 days (validation cohort, 70 patients) on ECMO and evaluated the associations between daily PaCO2, pH, and bicarbonate (HCO3) with sedation. Median PaCO2 was 41 (interquartile range [IQR] = 37-46) mm Hg and 41 (IQR = 37-45) mm Hg in the discovery and the validation cohort, respectively. Lower PaCO2 and higher pH but not bicarbonate (HCO3) served as significant predictors for reaching a Richmond Agitation Sedation Scale (RASS) target range of -2 to +1 (lightly sedated to restless). After multivariable adjustment for mortality, tracheostomy, prone positioning, vasoactive inotropic score, Simplified Acute Physiology Score (SAPS) II or Sequential Organ Failure Assessment (SOFA) Score and day on ECMO, only PaCO2 remained significantly associated with the RASS target range (adjusted odds ratio 1.1 [95% confidence interval (CI) = 1.01-1.21], p = 0.032 and 1.29 [95% CI = 1.1-1.51], p = 0.001 per mm Hg decrease in PaCO2 for the discovery and the validation cohort, respectively). A PaCO2 ≤40 mm Hg, as determined by the concordance probability method, was associated with a significantly increased probability of a sedation level within the RASS target range in both patient cohorts (adjusted odds ratio = 2.92 [95% CI = 1.17-7.24], p = 0.021 and 6.82 [95% CI = 1.50-31.0], p = 0.013 for the discovery and the validation cohort, respectively).