Pulmonary fibrosis leads to a decrease of oxygen diffusion, in particular during exercise. Bronchial obstruction also could decrease the partial pressure of oxygen (P(a)O(2)). In this study we investigated the validity of blood gas content, especially P(a)O(2) and P(a)O(2) affected by hyperventilation (P(a)O(2corr)) and alveolo-arterial oxygen gradient (P(A-a)O(2)) in comparison with the CO diffusion capacity (DLCO) in different lung diseases. A total of 250 subjects were studied (52.3 ± 12.5 year; F/M 40/210), among which there were 162 subjects with different lung disorders and 88 healthy controls. Pearson's correlation coefficients (r) of DLCO with P(a)O(2), P(a)O(2corr), and PA-aO(2) were analyzed in each group. The results show that the diagnostic power of P(A-a)O(2) against P(a)O(2corr) was equivalent, especially during exercise (r = -0.89 and -0.92, respectively). DLCO showed only weak correlations with P(a)O(2corr) and P(A-a)O(2) (r = 0.17 and -0.19, respectively). In conclusion, DLCO shows a better match with blood gas content during exercise than at rest during which it is routinely tested. Thus, the exercise test is advisable. The P(A-a)O(2) takes into account the level of ventilation, which makes it correlate better with DLCO rather than with blood gas content. The most significant problems in clinical evaluation of blood gas parameters during exercise are the insufficiently defined limits of normal-to-pathological range.