The transmission-disequilibrium (TD) test is a powerful method for det
ecting linkage between marker and disease loci in the presence of link
age disequilibrium. For multiallelic markers, we propose the use of ex
act tests, which are implemented using both an exact algorithm and Mar
kov chain Monte Carlo simulation. Simulation studies show that exact t
ests improve both the small sample validity and the power of the TD me
thod. We also compared the usual single-affected-offspring sampling sc
heme to one in which pairs of affected siblings are sampled. Affected-
sib-pair sampling greatly increases the power of the TD method and wil
l be most useful when a sample of affected sib pairs is available from
prior linkage studies. (C) 1997 Wiley-Liss, Inc.