A cross-sectional population survey of dynamic ventilatory lung function was performed to determine the variability of and reference values for time domain spirogram indices. The reference population comprised 83 men and 143 women who were healthy nonsmokers. The time domain indices were derived from the blow with the largest sum of FVC and FEV1. Prediction equations for FEV1 and mean flow between 25 and 75% FVC were little influenced by methods for selecting the data from the 3 blows recorded. The variability of the first 2 moments of the spirogram increased with age, requiring a log transformation for regression analysis. Prediction equations for conventional and time domain spirometric indices are presented and reference values are proposed based on an estimated percentile derived from the predicted value and the residual standard deviation from regression. Two time domain indices are identified that may prove better suited than conventional indices for identifying abnormalities in the tail of the spirogram.