The criterion-oriented problem of test bias and fairness in selection is compared to the construct-oriented problem of comparability of test scores in cross-cultural research. These problems are shown to have important similarities, and their studies may supplement each other. In a formulation of psychometric criteria for the comparability of test scores, three levels are described: functional equivalence, score equivalence, and item equivalence. Patterns of item responses are considered to be useful for establishing the similarity of an individual to a group. Recent attitudes toward test fairness include a shift from criterion orientation to decision orientation, and growing doubt about the possibility of finding a generally acceptable objective definition of fairness. An experiment with two ethnic groups in Surinam using 12 tests and four background variables showed fewer differences related to ethnic group than to sex or to socioeconomic differences. The selection procedures for secondary school students in Tanzania are briefly compared to current models, and bibliographical references are appended. (CTM)