Using the single breath test for carbon dioxide (SBT-CO2), the components of physiological deadspace were investigated during anaesthesia with IPPV in 58 patients. A square-wave inspiratory flow and an end-inspiratory pause (25% and 10% of cycle time, respectively) were used. At tidal volumes of 0.45 litre (f = 17 b.p.m.), and 0.75 litre (f = 9 b.p.m.), median values for VDphys/VT were 0.44 and 0.31. Increasing VT and decreasing f did not change airway deadspace (VDaw) so that the fraction VDaw/VT was decreased (P less than 0.001). The alveolar deadspace fraction, VDalv/VTalv, was decreased in 93% of patients (P less than 0.001). These improvements with increasing VT can be attributed to beneficial effects on gas distribution and diffusion time. Patients with large alveolar deadspaces had steeply sloping SBT-CO2 phase III, and increased expiratory time constants of the respiratory system. The median arterial--end-tidal PCO2 difference, (PaCO2-PE'CO2), was 0.6 kPa at small and 0.3 kPa at large tidal volumes (P less than 0.001). Three patients had zero and four had negative (PaCO2-PE'CO2) values at large tidal volumes. When phase III slopes steeply, negative (PaCO2-PE'CO2) values may be observed in the presence of alveolar deadspace.