Caseins are essentially concentrated in the colloidal fraction of ruminant milks as highly hydrated and mineralized spherical particles, termed casein micelles. They form a group of four peptide chains (alpha(s1), beta, alpha(s2) and kappa), encoded by four structural genes (CSN1S1, CSN2, CSN1S2 and CSN3, respectively) of which the expression is regulated by lactogenic hormones. These phosphoproteins are synthesized, essentially during lactation, in the mammary epithelial cells and we show, for the first time, that their regulation is also controlled at the translational level. Apparently, the four casein messenger are not translated with the same efficiency. Specific amplification systems have been developed and optimized to quantify, by real time quantitative PCR (qPCR), transcripts encoding the four caseins starting from total RNA extracted from mammary tissues taken on goats (n = 4), ewes (n = 3) and cows (n = 3), in lactation. The relative proportions of each specific messenger (% of casein mRNA) were compared to the relative amounts of the corresponding caseins (% of whole casein) in milks sampled from the same animals, determined after fractionation by reverse phase HPLC and integration of the corresponding peak areas. From qPCR data, the four casein transcripts appeared to be present approximately at the same level of abundance (ca. 25%, except for defective genotypes at the CSN1S1 locus, in the goat) whereas the amounts of the corresponding proteins in milk were ranging between 9 and 38% of the whole casein fraction. A comparison of specific translational efficiencies (% of protein in milk/% of transcript in the mammary tissue), showed that alpha(s1)- and beta-casein transcripts are translated ca. 3- to 4-fold more efficiently than alpha(s2)- and kappa-casein transcripts. This seems to be the rule in the three ruminant species studied. More or less optimal contexts for initiation of translation (Kozak recognition sequence of the start codon) as well as 3' untranslated region (UTR) sequences and length might explain, at least in part, our results. These preliminary results which have now to be confirmed with a larger number of individuals to strengthen our findings and conclusions, provides, however, a rational explanation to the unbalanced casein distribution (approximate proportions 4:1:4:1 for alpha(s1):alpha (s2):beta:kappa, respectively) reported for ruminant milks. The possible effects of specific secondary structures in the 5' and 3' UTRs of casein messengers still have to be considered.