In the presence of an externally applied thoracic restriction, conflicting ventilatory responses to exercise have been reported, which could be accounted for by differences in exercise protocol. Seven male subjects performed two incremental and two constant-workload ergometer tests either unrestricted or in the presence of an inelastic corset. Ventilatory variables and arterial estimates of PCO(2) were obtained breath by breath. Subjects hyperventilated in the presence of restriction during the constant-workload test (38.4 +/- 3.0 vs. 32.8 +/- 3.0 l/min for the average of the last 3 min of exercise, P < 0.05), whereas, at an equivalent workload during the incremental test, ventilation was similar to unrestricted values (unrestricted = 26.3 +/- 1.6 vs. restricted = 27.9 +/- 2.3 l/min, P = 0.36). We used a first-order linear model to describe the effects of change in workload on minute ventilation (24). When the time constants and minute ventilation values measured during unrestricted and restricted constant-workload exercise were used to predict the ventilatory response to the respective incremental exercise tests, no significant difference was observed. This suggests that hyperventilation is not seen in the restricted incremental test because the temporal dynamics of the ventilatory response are altered.