The aim of this study was to identify and rank the various factors, in particular those involving feeding, that affect the proportion of caseins in milk true protein. Twenty-nine feeding trials involving 821 lactations were assessed, and lactoprotein genetic variants were known for 551 of these. The main factor affecting the casein: protein ratio was the genetic variant of β-lactoglobulin: once corrected for other factors, the milk of BB type animals had a ratio nearly 30 g/kg total protein higher than AA animals. κ-Casein variant B also had a positive effect (+12 g/kg in favour of BB relative to AA animals). Except in the last weeks of pregnancy and the first weeks of lactation, the casein[ratio ]protein ratio varied little during lactation. It was significantly reduced when milk cell count exceeded 200000 cells/ml, even in the absence of clinical mastitis. It also decreased slightly with parity. Among the various dietary factors studied (level and type of nitrogen and energy supplies, forage type and preservation method), none had any significant effect on the milk casein[ratio ]protein ratio, except in drastic dietary situations. That ratio increased very slightly in parallel with the animals' milk yield and milk protein content. In practice, measuring the milk protein content in animals free from clinical mastitis remains a very precise predictor of casein content, accounting for 93% of its variation.