The multiple regression equation predicting lung function values for a study population of South African Blacks is compared with equations predicting normal values for Blacks elsewhere, and in almost all cases is found to predict higher values. This is so despite the study population's high prevalence of respiratory disease and long history of exposure to crocidolite asbestos dust. This anomalous finding is explored in terms of some problems with studies generating normal values. In particular, the confounding effect of social class status on ethnic determinants of lung function is considered. Low 'normal' values for Blacks reported from South Africa and elsewhere are examined. The disadvantages to workers of being evaluated in relation to low norms are discussed in terms of preventive medicine and workmen's compensation. The application of a universal standard for all is proposed.