Recent studies have shown that effective pulmonary ventilation is possible with tidal volumes (VT) less than the anatomic dead-space if the oscillatory frequency (f) is sufficiently large. We systematically studied the effect on pulmonary CO2 elimination (VCO2) of varying f (2-30 Hz) and VT (1-7 ml/kg) as well as lung volume (VL) in 13 anesthetized, paralyzed dogs in order to examine the contribution of those variables that are thought to be important in determining gas exchange by high frequency ventilation. All experiments were performed when the alveolar PCO2 was 40 +/- 1.5 mm Hg. In all studies, VCO2 increased monotonically with f at constant VT. We quantitated the effects of f and VT on VCO2 by using the dimensionless equation VCO2/VOSC = a(VT/VTo)b(f/fo)c where: VOSC = f X VT, VTo = mean VT, fo = mean f and a, b, c, are constants obtained by multiple regression. The mean values of a, b, and c for all dogs were 2.12 X 10(-3), 0.49, and 0.08, respectively. The most important variable in determining VCO2 was VOSC; however, there was considerable variability among dogs in the independent effect of VT and f on VCO2, with a doubling of VT at a constant VOSC causing changes in VCO2 ranging from -13 to +110% (mean = +35%). Increasing VL from functional residual capacity (FRC) to the lung volume at an airway opening minus body surface pressure of 25 cm H2O had no significant effect on VCO2.