Spirometric parameters were measured in 300 healthy, lifetime nonsmoking American Indians. When "best" is defined by the highest correlation coefficient and smallest residual standard deviation, the best prediction equations were linear regression equations using only age and height as the independent variables. Visual comparisons of two-dimensional graphic representations of each predictor in these equations with equations commonly used to predict spirometric variables in white persons revealed no substantial differences. However, statistical comparisons, using an analysis of covariance, with data from a previous study of white persons in our laboratory showed the forced vital capacity and forced expiratory volume in 1 second equations for Indian men to be different from the equations for white persons. No statistically significant differences were found between the prediction equations for Indian and white women. The meaning of the statistically significant differences is unclear and they may not have any clinical relevance. Until this issue is resolved we recommend that race-specific equations be used when possible.