Because applicant populations for many jobs and educational opportunities are becoming increasingly multicultural, the whole issue of fairness in testing has become highly salient. In South Africa, a main reason for the poor performance of the economy is the failure to make optimal use of all manpower, one of the crippling legacies of the apartheid era. There is also a strong moral motivation on the part of some employers to do something about the inequalities of the past by advancing greater numbers of disadvantaged individuals to skilled jobs. The question is one of how to identify the "right" individuals from disadvantaged backgrounds for these opportunities, since there are drawbacks to relying on conventional test scores to make selection, training, and advancement decisions. Three approaches to this problem are discussed in this paper: (1) putting in place of fairness programs; (2) use of trainability tests to make selection decisions; and (3) use of learning potential measures to make such decisions. Although practical learning potential measures suitable for application in industry do not yet exist, such measures are likely to be a useful tool. (SLD)