Abstract: PCR-RFLP was used to genotype 87 Sarabi native cattle of north-western Iran for A and B alleles of kappa casein gene. A 350 bp length of exon 4 and intron 4 was amplified and digested with HinfI endonuclease. Samples were loaded on agarose gel (2%) and genotyped under UV light. Allele frequency of desirable B allele was 0.57. Stochastic simulation was used to generate milk yield trait for a population of 4950 females and 50 males for 15 overlapping generations. Population parameters included 1100 and 436 kg for average milk yield and phenotypic deviation, respectively; with heritability of 0.27. Additive and dominance effects of Kappa Casein gene were considered as 187.63 and 50.37 kg, respectively. Two methods were considered for selection of males based on the first phenotypic record of their dams (PAS) or molecular information of each male, individually (GAS). Females were always selected on their first phenotypic record. Although, there was a significant difference between polygenic and major gene genetic response between two methods after the 5th generations, but there was no significant difference for the sum of polygenic and major gene response. After 15 generations of selection there was no significant difference between inbreeding coefficient under two methods. Selection plan for males based on one single major gene had no advantage over the conventional selection based on dam record in native Sarabi breed.