1. The intensity of breathlessness in normal subjects during exercise was measured on seven occasions over a 40-week study period to assess the long-term repeatability of Borg scale estimates of breathlessness. 2. In all subjects there was a significant correlation (P = 0.0001) between breathlessness and minute ventilation. Minute ventilation measured at each work rate did not differ between the seven exercise tests (P greater than 0.05). 3. There was no significant difference between the mean Borg scores (measured with respect to a given level of ventilation) in 5 of the 7 test weeks (P greater than 0.05). The slope of the relationship Borg score/minute ventilation did not differ between the seven exercise tests (P greater than 0.05). 4. Breathlessness estimation was highly reproducible both after 1 week and after 40 weeks of the study (both P greater than 0.05). 5. The duration without testing between consecutive tests did not affect reproducibility: the mean Borg score was as equally reproducible after an interval of 22 weeks without testing as after an interval of 1 week (P greater than 0.05). 6. The Borg scale provides a reliable technique for studying the sensation of breathlessness over extended time periods.