During arm elevation, normal individuals predominantly recruit the diaphragm, whereas chronic obstructive pulmonary disease (COPD) patients use more the accessory inspiratory and abdominal expiratory muscles. To test that arm elevation is useful to study the ventilatory muscle response in COPD, and to define the factors that determine this response, we studied 34 patients (FEV1 0.95 +/- 0.08 L) during 2 min of arm elevation. Transnasal balloons were used to determine end-inspiratory and end-expiratory gastric (Pg), pleural (Ppl) and transdiaphragmatic (Pdi) pressures. The slope of delta Pg/delta Ppl (= Pgi - Pge/Ppli - Pple) was used to infer respiratory muscle recruitment. With linear regression, the delta Pg/delta Ppl during arm elevation significantly correlated with resting delta Pg/delta Ppl (r = 0.68), hyperinflation (FRC/TLC, r = 0.52), and diaphragmatic tension time index (TTIdi) (r = 0.47). With multiple regression, resting delta Pg/delta Ppl, percentage predicted FRC (FRC %pred) and TTIdi influenced delta Pg/delta Ppl during arm elevation (r = 0.84). Over the 2 min of arm elevation, the dependency upon resting breathing pattern decreased, while hyperinflation and TTIdi became increasingly important. Higher resting TTIdi values were associated with a faster and shallower breathing pattern (f/VT > or = 38) during arm elevation. We conclude that the pattern of respiratory muscle recruitment during arm elevation depends primarily on the resting breathing pattern. Over 2 min, the degree of hyperinflation and the force reserve of the diaphragm increasingly impact on the ability to recruit the diaphragm. Measurement of f/VT during arm elevation is useful to determine functional reserve of the diaphragm in severe COPD.