The transmission/disequilibrium test (TDT) was introduced several years ago by Spielman et al.(1993) as a test for linkage between a complex disease and a genetic marker. The original intended use of the TDT was to test for linkage with a marker located near a candidate gene, in cases where disease association already had been found. However, even if prior evidence for association is absent, the TDT is valid and can be used to test any marker (or a set of markers) for which data are available from parents and one or more affected offspring. Other tests that focus on association itself-but that, like the TDT, are applied to data from nuclear families-also have been proposed. In addition, several papers have appeared recently that discuss extensions and properties of these tests and of the TDT. Since the aim of both the TDT and the association tests is to locate genes that contribute to disease susceptibility, a number of questions about their differences have arisen. In this review, we compare the properties of the TDT with those of family-based tests of association, and we comment on issues regarding the use of these tests.

The Model and the Problem The genetic model consists of both a locus D that contributes to disease susceptibility and a (possibly linked) marker locus M. The standard way to identify disease loci is to use classical (LOD) or nonparametric (affected-sib-pair [ASP]) methodsto test for linkage with such a marker or a set of markers. It has been known for some time, however, that these methods may fail to detect a disease locus linked to a marker, even though the locus may be of biological significance (Cox et al. 1988; Spielman et al. 1989). Even when standard linkage tests fail to provide evi-dence, however, a disease locus linked to the marker