Our predicted equations were derived by aiming at minimizing the absolute difference between the observed and the predicted values for each spirometric parameter (FEV1 and FVC), rather than that of their predicted percentages (percentage of predicted FEV1 or percentage of predicted FVC = [observed FEV1 or FVC/predicted FEV1 or FVC] × 100%). With regard to the predicted percentages, we compare the use of the three prediction equations using our original set of 595 Hong Kong women with mean age of 46 years (SD, 17 years; range, 10 to 80 years) and mean height of 155 cm (SD, 6.3; range, 133 to 183). There is no evidence of outliers, and normality assumption is checked and fulfilled. The 95% confidence intervals for percentage of predicted FEV1 using the three sets of equations of Ip et al,1Knudson et al,2and National Health and Nutrition Examination Survey3 are 98.8 to 101.2, 104.9 to 107.5, and 83.3 to 85.2, respectively, while the corresponding 95% confidence intervals for percentage of predicted FVC are 98.8 to 101.2, 84.3 to 86.4, and 58.3 to 61.1. As a result, our equations give the most accurate predicted percentages, with the corresponding confidence intervals including 100, and in effect outperform other nonlocal equations when applied to our local sample.