Graduate admissions tests are used to choose pupils for postbaccalaureate academic or professional degree programs. Their use is mainly justified by a demonstration of predictive validity--the extent to which they predict some important criterion or outcome. A situation in which an examination is used to predict a specific criterion for a particular population, and is found to give systematically different predictions for subgroups of the population who are identical on that that specific criterion, is called "predictive bias." Fairness to the group versus fairness to the individual is discussed, as are internal and external methods of detecting bias. Some external methods of detecting bias include differential validity coefficients, and differential prediction systems. Limitations of the differential validity approach are discussed. (SGK) (53pp.)