Casein αs1 gene has a major effect on protein content of dairy goat milk. All French bucks (Alpine and Saanen breeds) used for artificial insemination, 2 145 testing bucks and 2 983 dams of these bucks, have been genotyped for this gene since 1990. The casein αs1 gene is a multi-allelic gene with 6 different alleles. The idea of this study is to investigate how to include the casein αs1 gene effect in genetic or genomic evaluation. In French dairy goat, Casein αs1 genotype has a significant effect on protein content, milk yield and fat content. Parts of variances of casein αs1 genotype were estimated between 3% in milk yield in Saanen breed to 38% for protein content in Alpine breed. Genetic evaluations based on daughter yield deviations (DYD) were done per breed, including casein αs1 genotype as fixed or random effect. Validation correlations estimated on the 252 youngest bucks were slightly improved by considering the casein αs1 gene in Saanen breed (+18% in protein content). However considering the gene effect as fixed or random gave similar results. Adding the 50k bead chip genotypes of bucks (471 Alpine and 354 Saanen bucks) in the model already including casein αs1 genotype did not improved validation correlation. The most of females used for genetic evaluation were not genotyped for casein αs1 gene. Including casein αs1 gene effect in genetic evaluation based on female performances need to predict casein αs1 genotype for these females. Probabilities of each 19 genotypes were estimated for each female, from pedigree information and true genotype of animals. Several models were tested to include this probability in the model as random or fixed effect. The first results shown that validation correlations were not improved considering the genotype probability. Other methods or models are investigated to avoid the problem of imputation.