1. To test the hypothesis that repeated associations of exercise and increased respiratory dead space elicit mechanisms that augment future ventilatory responses to exercise alone, experiments were conducted on normal adult goats familiarized with experimental procedures. 2. Measurements of ventilation, arterial blood gases and CO2 production were made at rest, during mild steady-state exercise (4 km h-1; 5% grade) and with increased dead space at rest in seven goats before and after training. In Series I experiments, training consisted of fourteen to twenty exercise trials explicitly paired with increased dead space (0.8 l) over 2 days. Increased dead space predominantly represents a CO2 chemoreceptor stimulus with only mild hypoxic stimulation. Post-training measurements were made 1-6 h and 1 week after training was completed. 3. The same goats repeated a slightly modified protocol several months later (Series II; 6 trials per day for 4 days) to determine if responses were both repeatable and reversible, and to investigate training effects on dynamic ventilatory responses at the onset of exercise. 4. In Series I experiments, resting minute ventilation and breathing frequency were elevated 1-6 h post-training compared to baseline (44 and 74% respectively), whereas resting tidal volume decreased (14%). One week post-training, resting values had returned to baseline. Series II training had no significant effects on resting measurements. 5. Relative to baseline, arterial partial pressure of CO2 (Pa,CO2) values decreased significantly more from rest to exercise 1-6 h post-training in both Series I (2.7 +/- 0.2 vs. 1.8 +/- 0.9 mmHg) and Series II (3.4 +/- 0.6 vs. 2.0 +/- 0.6 mmHg). The exercise ventilatory response increased 25-28% 1-6 h post-training (both series), largely due to a greater exercise frequency response, but returned to baseline 1 week post-training. Training had no effect on ventilatory responses to CO2 at rest, suggesting that decreases in CO2 chemoreceptor responsiveness did not cause the greater exercise ventilatory response. Model estimates indicate that the net feedforward exercise ventilatory stimulus was increased 40-50% by training. 6. Training had no discernable effects on ventilatory dynamics at the onset of exercise. However, post-training differences in Pa,CO2 regulation and ventilation were established early in exercise, prior to steady state. 7. Collectively, these experiments suggest a previously unsuspected degree of repeatable and reversible plasticity in the control system subserving the exercise ventilatory response. Such plasticity may contribute to the development of normal exercise hyperpnoea and to adaptive responses of the ventilatory control system in adult animals.