In healthy subjects, we compared the effects of an expiratory (ERL) and an inspiratory (IRL) resistive load (6 cmH2O.l-1.s) with no added resistive load on the pattern of respiratory muscle recruitment during exercise. Fifteen male subjects performed three exercise tests at 40% of maximum O2 uptake: 1) with no-added-resistive load (control), 2) with ERL, and 3) with IRL. In all subjects, we measured breathing pattern and mouth occlusion pressure (P0.1) from the 3rd min of exercise, in 10 subjects O2 uptake (VO2), CO2 output (VCO2), and respiratory exchange ratio (R), and in 5 subjects we measured gastric (Pga), pleural (Ppl), and transdiaphragmatic (Pdi) pressures. Both ERL and IRL induced a high increase of P0.1 and a decrease of minute ventilation. ERL induced a prolongation of expiratory time with a reduction of inspiratory time (TI), mean expiratory flow, and ratio of inspiratory to total time of the respiratory cycle (TI/TT). IRL induced a prolongation of TI with a decrease of mean inspiratory flow and an increase of tidal volume and TI/TT. With ERL, in two subjects, Pga increased and Ppl decreased more during inspiration than during control suggesting that the diaphragm was the most active muscle. In one subject, the increases of Ppl and Pga were weak; thus Pdi increased very little. In the two other subjects, Ppl decreased more during inspiration but Pga also decreased, leading to a decrease of Pdi. This suggests a recruitment of abdominal muscles during expiration and of accessory and intercostal muscles during inspiration. With IRL, in all subjects, Ppl again decreased more, Pga began to decrease until 40% of TI and then increased.(ABSTRACT TRUNCATED AT 250 WORDS)