Changes in gas viscosity during the nitrogen washout test will affect pneumotachograph measurements. A digital correction technique, using estimates of viscosity derived from measurements of gas concentration, has been described and was examined in this study. The pneumotachograph gain varied directly with argon concentration (coefficient of correlation = 0.995, P less than 0.01) indicating that a simple correction was legitimate. This was tested with an in vitro model, giving very consistent estimates of volume (FRC) despite variations in the ventilation pattern, which were slightly underestimated due to incomplete mixing in the system. Alveolar mixing efficiency (AME) was also reproducible but, like studies in man, fell when low tidal volumes were used. In a repeatability study in six volunteers, examined six times on each of two days, FRC and AME were also reproducible (average coefficient of variation within subjects of 4.97% and 4.34% respectively), considerably better than another reported study using conventional flow measurements. This method appears to be a useful addition to existing techniques for studying nitrogen washout.