If most agree with the dual pattern/process definition of adaptation, an apomorphy promoted by natural selection, few indeed identify an adaptation using the same methodological procedure. Most differences between procedures rest with the way the role of natural selection is analysed in a phylogenetic perspective. We argue that the selective value of a character cannot be phylogenetically reconstructed, but must be considered independently from the phylogenetic analysis in the framework of population biology. The phylogenetic pattern for the character of interest can only be used to refute or corroborate circumstantially the macro-evolutionary predictions issuing from population studies.