The purpose of this study was to establish the proton NMR relaxation times of collapsed but otherwise normal lung tissue and to determine whether an inflammatory process within a collapsed lung can be detected by alterations in relaxation times. The lungs of three groups of rabbits were studied: group A (n = 7) had a sterile collapse of one lung for two days. The two other groups also had one lung collapsed, but with bacterial (group B, n = 6) or chemically induced (group C; n = 6) pneumonitis superimposed. The contralateral lung, which was acutely deflated at the time of thoracotomy, served as a control in each animal. T1, T2 and the total water content were measured on freshly excised lung samples. In group A, there was no significant difference in T1 (606 +/- 14* ms vs. 595 +/- 18 ms;* = SEM) or T2 (80.6 +/- 1.7 ms vs. 78.4 +/- 2.6 ms) between the collapsed and the control lung tissue. In each animal in groups B and C, T2 was longer in the collapsed lung with superimposed pneumonitis than in the control lung tissue (group B: 116.8 +/- 6.9 ms vs. 82.9 +/- 1.8 ms, P less than .001; group C: 120.5 +/- 5.9 ms vs. 86.0 +/- 1.5 ms, P less than .001). T1 changes were similar, but less marked. There was a linear relationship between the relaxation times and the total water content of the lung samples (T1:r = 0.87; T2:r = 0.91). It is concluded that proton NMR may have a potential in detecting disease such as inflammation in collapsed lung tissue based on differences in relaxation parameters compared with normal lung areas.