The aim of this prospective study was to analyse the contribution of the measurement of alveolar arterial gradients of CO2 during forced expiration in the diagnosis of pulmonary emboli occurring in chronic airflow obstruction (COPD) as a result of smoking. The study was carried out on 178 patients: Group 1: 54 subjects without emboli (14 controls, 33 COPD and 7 patients with chest pain); Group 2: 72 patients with proved emboli (49 non COPD, 23 COPD); Group 3: 52 patients COPD presenting with varied non-embolic broncho-pulmonary pathology (pneumonia, bronchospasm, pulmonary oedema, bronchial neoplasm). The diagnosis of pulmonary emboli was confirmed by scintigraphy in patients with non COPD or angiography (in patients with COPD). The maximal fraction of CO2 was measured using a capnologue during a forced expiration which was long and prolonged until residual volume was achieved. The PaCO2 was measured simultaneously by an analysis of arterial blood gases. The D index was calculated according to the formula [(PaCO2-PEM CO2)/PaCO2] x 100. The D index was significantly lower in Group 1 (3.42 +/- 3.8% p < 0.0001) than in Group 2 (20.8 +/- 10%) and Group 3 (17.6 +/- 11.7%) (not significant between Groups 2 and 3). In patients with COPD the specificity and sensitivity and the predicted positive and negative value were 100% for a D limit of 7%. In COPD patients these values were respectively 82, 95, 75 and 96% for a D limit of 7%; on the other hand for a D below 5% the values were 60, 100, 64 and 100% respectively.(ABSTRACT TRUNCATED AT 250 WORDS)