1. The effect of residence at altitude on the perception of breathlessness after return to sea level was examined in normal subjects. Breathlessness (Borg scale), minute ventilation, respiratory frequency, tidal volume, 'oxygen pulse' (oxygen consumption/heart rate) and the ventilatory equivalent for oxygen (minute ventilation/oxygen consumption) were measured at exercise (cycle-ergometer) during 5 months of training before 4 weeks at 4000 m and during the 6 month period after return to sea level. 2. There was no change in the subjects' pattern of breathing (respiratory frequency and tidal volume) or 'oxygen pulse' after the period at altitude (P = 0.0001). The ventilatory equivalent for oxygen was increased at all work rates after the period at altitude (P = 0.02). This ratio was slightly lower after 6 weeks and had returned to normal by 6 months (P = 0.4). 3. During training there was no change in breathlessness score (P = 0.6). On return to sea level, breathlessness score relative to ventilation was reduced (P = 0.0001). This was maintained for at least 6 weeks, but not as long as 6 months. 4. This study has demonstrated that, in normal subjects, the otherwise stable and reproducible relationship between breathlessness and ventilation may be disrupted for several weeks by factors other than lung disease. 5. The mechanism responsible for this is not clear, but the observations are consistent with the hypothesis that prior experience of breathlessness may condition subsequent estimates of breathlessness.