This study covered two main areas of major genes affecting quantitative traits: (i) the estimation of their effects with emphasis on the milk protein loci and (ii) the use of genotype information on major genes as part of the selection criteria. In a situation in which only a subset of the population has known genotypes for a major gene, the estimated effects of this gene obtained with a method using performance information on all the individuals (with and without known genotype) were compared with those estimates obtained with a method using information on only individuals with known genotype. The first method used a Gibbs sampling approach to infer genotypes of individuals with unknown value. The results from a simulation study showed that, in absence of selection, both methods yielded unbiased estimates of the major gene effects. However, the inclusion of performance information of individuals without genotype decreased the error variance of the estimates by 12 to 69% of the reduction there would be if all individuals had known genotype, depending on the gene frequency, and the mode of action of the major locus. In the population undergoing selection the use of such information also substantially reduced the bias of estimates. This methodology was applied to estimation of the effects of the β-lactoglobulin and the κ-casein loci on lactation traits (milk yield, fat and protein yield and content), using data from 1452 Holstein Friesian cows of two experimental herds and a MOET nucleus in the UK, and available progeny test of sires. There were no significant effects of these loci on any of the traits considered. To study the use of genotype information as part of the selection criteria, a deterministic model for predicting response to selection when a single locus is segregating was defined. It was used to compare the traditional phenotypic selection with other methods of combining performance information with either the genotype of the major locus or only its Mendelian sampling term (i.e. the effect due to the major locus expressed as deviation from family mean).