1. The intensity of breathlessness was measured during exercise in nine normal subjects using a modified Borg scale to examine the effect of prior experience of breathlessness on subsequent estimates of breathlessness. 2. Each subject performed four exercise tests, each of which consisted of two identical runs of workload incrementation (run 1 and run 2). An inspiratory resistive load of 3.8 cmH2O s-1 l-1 was applied during the appropriate run of the exercise test to examine the effect of (a) prior experience of 'loaded' breathing on breathlessness estimation during 'unloaded' breathing, and (b) prior experience of 'unloaded' breathing on breathlessness estimation during 'loaded' breathing. Run 1 was the conditioning run: run 2 was the run in which the effect of conditioning was measured. 3. There was a good correlation between breathlessness and minute ventilation during both 'unloaded' breathing (median r = 0.93) and 'loaded' breathing (median r = 0.95). 4. The slope of the Borg score/minute ventilation relationship was greater during 'loaded' breathing than during 'unloaded' breathing (P less than 0.01). There was no difference in mean Borg score between 'unloaded' and 'loaded' breathing. 5. After a period of 'loaded' breathing during run 1, estimated breathlessness was significantly reduced during ensuing 'unloaded' breathing in run 2 (P less than 0.01) compared with the exercise test in which 'unloaded' breathing was experienced throughout both run 1 and run 2. 6. After a period of 'unloaded' breathing in run 1, estimated breathlessness was significantly increased during ensuing 'loaded' breathing in run 2 (P less than 0.01) compared with the exercise test in which the inspiratory load had already been experienced in run 1.(ABSTRACT TRUNCATED AT 250 WORDS)