Abstract： Genetic studies of quantitative traits can be accomplished through the method of selective genotyping. In this article, we proposed a statistic T for association analysis of a quantitative trait locus (QTL) by selective genotyping. The statistic T is to compare the difference between the mean trait values for homozygotes of the marker in the upper extreme sample. To evaluate the statistical properties of the statistic T , its distribution and type I error rate were examined under the null hypothesis of no association by simulation. Simulation results showed that the distributions of T were similar to the theoretical central x~2 distributions and the type I error rates were around the nominal significance levels for various sample-selection strategies. The power of the statistic T was investigated under various heritability, various sample sizes, and various percentiles selected for three genetic models. The results showed that the power always increases with increasing linkage disequilibrium (LD) between the marker and the trait locus. As expected, power increases when sample size and heritability increase, and also increases with more-stringent selective-genotyping procedures.