The respective value of four non-invasive methods for the diagnosis of pulmonary arterial hypertension (PAH) was investigated in 63 COPD patients, using right heart catheterization as the reference method: 22 patients had no resting PAH (pulmonary artery mean pressure (PAP) less than or equal to 20 mmHg); 26 patients had mild PAH (PAP = 21-30 mmHg); and 15 patients had moderate to severe PAH (PAP greater than 30 mmHg). The specificity of ECG was 86% and the sensitivity 51% (only 38% in mild PAH). The specificity of radiological measurements was 63% and the sensitivity 46% (38% in mild PAH). Echocardiography (echo) had the best results with a specificity of 75% and a sensitivity of 78% (73% in mild PAH), but reliable echo measurements were available in only 52 out of 63 patients. Myocardial scintigraphy had a specificity of 68% and a sensitivity of 66% (58% in mild PAH). A stepwise regression analysis (including one echo, one ECG, one radiological and one functional variable) explained only 43% of the variance of PAP (multiple r = 0.66). These results suggest that no individual method is sufficiently reliable for predicting the presence of PAH, and particularly mild PAH, but the combination of echo + myocardial scintigraphy allows the prediction of PAH with a good probability. The precise level of PAH cannot be estimated, even when using multiple regression equations.