A detailed analysis is presented of the relationship between genetic markers and quantitative trait loci (QTLs) in the process of marker-assisted selection (MAS). We simulated MAS employing a multiple linear regression to chose from among all of the markers in the genome those to be utilized by selection and to estimate their associated effects on the trait. The simulations demonstrate that, even when such selection is quite effective, the markers utilized by selection are not necessarily the most tightly linked to the QTLs controling the trait. Moreover, the additive effects associated with the markers estimated by the regression may not accurately reflect the contributions to the trait by the most tightly linked QTLs.