Results of practical tests with whole milk and the three most allergenic proteins were compared in 32 children who were suspected of showing allergy to milk. Diagnosis was composed of a suggestive clinical history, familial atopy, positive in vitro and in vivo tests and regression of symptoms under an exclusion diet. Tests of re-introduction were done with 11 children. Better agreement between the provocation tests and the other elements of the diagnosis was found when the tests were made with the fractions, rather than whole milk. Furthermore, the finding of reduced allergy to casein allows consideration of re-introduction of milk in the form of cheese.