Previous studies have shown that Marker-Assisted Selection (MAS) based on a known major gene and genotypic selection can increase response to selection in the short term but reduce longer-term response to selection. The criterion for genotypic selection is:$$I = g + h\sp2(P - G)$$where g is the breeding value for the major gene and h\sp2(P−G) is the animal's polygenic EBV based on phenotype (p) adjusted for major genotype (G). Recently, Dekkers and van Arendonk (1998) developed methods to optimize the use of a known major gene in selection to maximize response over a planning horizon. The objective of this study was to use stochastic simulation to evaluate the optimal strategies developed by Dekkers and van Arendonk (1998) under a model in which genetic variance declined as a result of selection (Bulmer effect). A population with discrete generations, fixed size and equal selection among males and females was considered. Resulting strategies maximized cumulative response to selection for a pre-specified planning horizon. Results show that optimal strategies that are derived under a model with constant genetic variance may not result in greater responses to selection for all situations. Further improvements in optimal strategies are under development.