Rehabilitation programmes in chronic obstructive pulmonary disease (COPD) require exercise training above the anaerobic threshold. However, not all COPD patients develop metabolic acidosis during exercise. The hypothesis of this study was that non-exercise variables, characterizing the mechanical load on the inspiratory muscles during breathing at rest, can be used to reliably predict which patients with COPD are not able to develop metabolic acidosis during exercise. Thirty participants with COPD performed a symptom-limited cycle ergometer test. The oesophageal pressure/time index (PTIoes: the product of pressure magnitude and duration), the mean rate of pressure development during inspiration (Poes/TI), and the mean airway resistance (Raw)/maximal oesophageal pressure (Poesmax) ratio served as indices for the mechanical load on the inspiratory muscles. The oxygen uptake (VO2) at which plasma standard bicarbonate was seen to decrease from its baseline value was taken as the anaerboic threshold (AT). Mean Raw was significantly higher in those patients in whom the AT could not be detected. No other lung function parameters measured at rest allowed the accurate selection of those patients who did or did not develop exercise metabolic acidosis. On the other hand, Raw/Poesmax, PTIoes and Poes/TI were significantly different in the two patient groups. Additionally, whereas in the patient group with identifiable AT exercise hyperpnoea produced a non-linear increase of Poes/TI with respect to PTIoes above the AT, in the patient group without identifiable AT there was a linear relationship between Poes/TI and PTIoes throughout exercise. We conclude that the determination of inspiratory muscle load indices at rest may be useful in pulmonary rehabilitation programmes, for identifying those patients with COPD who do not develop exercise induced metabolic acidosis.(ABSTRACT TRUNCATED AT 250 WORDS)