Progressive threshold loading (PTL) is a common test of respiratory muscle endurance. Healthy naive subjects improve endurance with successive exposures to PTL by altering their breathing responses, thus necessitating a familiarization period before reproducible measures can be obtained. This study sought to determine whether a similar "learning effect" is evident in patients with COPD, and what the mechanism of any such effect may be. Ten subjects with COPD (FEV1 34+/-13% predicted) underwent PTL on four occasions (>24 h apart). During PTL measurements were obtained of breathing pattern and maximum threshold pressure (Pthmax) achieved. Maximum inspiratory pressure (PImax) was measured on each occasion. Over the four tests PImax improved by 21+/-16% (SD) (P<0.05) and Pthmax by 32+/-21% (P<0.05) with a plateau in these measures achieved by test three. Pthmax/PImax was unchanged, being 61+/-11% at test one and 67+/-12% at test four. In contrast to healthy subjects, PTL was not associated with increased expiratory time or decreased end-expiratory lung volume. In contrast to PImax and Pthmax, which changed with successive tests, a single measure of the ratio Pthmax/PImax may present a useful guide to the endurance capacity of the respiratory muscles in patients with COPD.