In chronic obstructive pulmonary disease (COPD) patients, there is a difference between PaCO2 and end-tidal partial pressure of CO2 (PetCO2). This gradient P(a-et)CO2 is due to ventilation/perfusion mismatching and deadspace, and is usually abolished by forced and prolonged expiration. We hypothesized that this gradient might not be canceled by forced expiration in the case of acute respiratory failure (ARF) related to pulmonary embolism (PE). Forty-four adult COPD patients were prospectively entered into this study; they were suspected of having ARF related to PE on the basis of clinical and biological data on admission. Maximum expired partial pressure of CO2 (PemCO2) was measured in mechanically ventilated and sedated patients by an interrupt of mechanical support. CO2 concentration was recorded during the following prolonged and passive expiration. The test was considered valid if an expiratory plateau was obtained. PemCO2 was measured in triplicate. Simultaneously, PaCO2 was measured and the ratio, R = [( 1-PemCO2]/PaCO2) x 100, was calculated. Pulmonary angiography was performed on the same day for all patients. Results showed that 17 patients had PE (PE+) and 17 had no PE (PE-). The two groups were comparable regarding mean age, severity of underlying chronic respiratory disease, PaCO2, PaO2, and hemodynamic data on admission. P(a-em)CO2 and R were significantly different in PE+ and PE- patients at 12 +/- 6.9 torr compared to 1 +/- 2.4 torr and at 28 +/- 14.8% compared to 2 +/- 6.2% (p less than .001), respectively. The positive predictive value of the test was 74%, but the negative predictive value 100% and the specificity was 65%, but sensitivity was 100%.(ABSTRACT TRUNCATED AT 250 WORDS)