The response of skeletal muscle to training is influenced by both the intensity and nature of the training stimulus. In this study we investigated the characteristics of the ventilatory load applied to the ventilatory muscles during several different modes of ventilatory muscle training. Patients with chronic obstructive pulmonary disease (COPD) performed the following breathing maneuvers: (1) Unloaded hyperpnea (UH), (2) resistive breathing through a fixed orifice (0.5 cm diameter) at frequencies of 15 and 30 breaths/min (RT15, RT30), (3) loaded breathing through a threshold valve set at 30% of the PImax at frequencies of 15 and 30 breaths/min (TT15, TT30), and (4) repetitive maximal inspiratory maneuvers against a closed shutter (PImax). During these maneuvers were recorded airflow and pressures at the month and esophagus, and from these measurements we derived VE and the work of breathing (WOB), tension time index (TTI), and pressure time product (PTP). The VE during UH was significantly higher than all other modes (p < 0.01), whereas the Pesmax was significantly lower during UH than during the resistive and loaded maneuvers (p < 0.01). The WOB did not differ during UH, TT30, and RT30, but was significantly higher in all three modes than at TT15 and RT15 (p < 0.05). During RT30 the TTI was higher than during TT30, TT15, and RT15 (p < 0.05), whereas the TTI during UH was significantly lower than during other maneuvers (p < 0.01). As expected, the highest Pesmax and PTP were found during the PImax maneuver. These data show that important qualitative differences in ventilatory muscle loading can be achieved by means of different devices and breathing strategies.(ABSTRACT TRUNCATED AT 250 WORDS)