1. Six normal subjects cycled to endurance or for 60 min at four work rates (WR 1-4): mean of 34% working capacity (93 watts for 60 min); 43% (120 watts for 56 min); 63% (177 watts for 37 min); and 84% (233 watts for 12 min), to determine how breathing pattern and dyspnoea change during prolonged activity. Four to six minutes were allowed to establish steady state and subsequent changes were considered to be endurance related. 2. Dyspnoea (Borg scale, 0-10) increased with the duration of activity at all work rates. 3. Ventilation (VE) did not change at WR1; increased from 44 to 47 l min-1 at WR2; from 60 to 88 l min-1 at WR3; and from 111 to 132 l min-1 at WR4. Dyspnoea was significantly and independently related to ventilation and duration of activity: dyspnoea = 0.004 VE1.36 time 0.25 (r = 0.81; partial F 202 and 26 respectively). 4. Inspiratory resistance did not increase at any work rate. Dynamic elastance remained constant during WR1, WR2 and WR3 but increased from 7.4 to 9.1 cmH2O l-1 during WR4. 5. Peak inspiratory pressure did not increase, and the increase in VE was accomplished by an increased breathing frequency without change in duty cycle. 6. Duration of activity is an important contributor to dyspnoea independent of changes in respiratory muscle contractile activity.