Simulation experiments were used to determine the empiric type I error rate of the Haseman-Elston sib-pair test for linkage between a quantitative trait and a marker locus in samples with 60 or fewer sib pairs. The effect of different levels of marker-locus heterozygosity on statistical validity was also considered. The test was performed on the trait and each of five unlinked markers, and evaluated using two different degrees of freedom for the t-distribution. The number of degrees of freedom in the first evaluation was based on the number of sib pairs in each sibship, sigma s(i)(s(i) - 1)/2 - 2. In the second evaluation, the number of degrees freedom was based on the number of sibs minus 1 in each sibship, sigma (s(i) - 1) - 2. Empirically determined type I error rates using sigma s(i)(s(i) - 1)/2 - 2 degrees of freedom were slightly liberal. For sigma (s(i) - 1) - 2 degrees of freedom, the estimated empiric p-values were nearly identical to their respective nominal p-values. Decreasing levels of heterozygosity did not increase the empiric type I error rate when the sample size was small.