The dynamic influences of end-tidal CO2 and exercise on ventilation are compared when CO2 and exercise are imposed separately and when they are imposed simultaneously. Five human subjects are studied. The subjects performed three trials: random work rate forcing, random CO2 inhalation and their simultaneous loading. The work rate was varied between 20 and 80 W as a pseudorandom binary sequence. The concentration of inspired CO2 was varied randomly between 0 and 7 per cent, adjusted so that it produced approximately the same amount of ventilatory fluctuations as the random work load. The relative contribution of each variable was analysed using multivariate autoregressive analysis at frequencies ranging from 0.1 to 1 cycle min-1. The results show that the dynamics of the response to CO2 inhalation, exercise and their combination are nonlinear and that the combination of CO2 inhalation and exercise magnifies the nonlinear behaviour. Ventilation is largely unaffected by either work rate or end-tidal CO2 at 1 cycle min-1. During simultaneous CO2 and work rate forcing, ventilation tends to follow the change in the end-tidal CO2.