Determining the direction of past transitions between adaptive traits is one of the major objectives of evolutionary biology. Insights can be gained from phylogenies, but violation of the assumptions of the statistical models used to reconstruct traits can result in severe biases and complementary evidence should be considered. Here, we review the weaknesses of relying solely on species phylogenies in reconstructing the evolutionary history of C(4) photosynthesis in grasses, a complex trait present in distinct phylogenetic groups. We argue that evolutionary transitions should be reconstructed by establishing the homology or convergence of the different states based on genetic and phenotypic analyses. Such an approach points to a predominance of C(4) gains over reversals to C(3) and we discuss potential explanations for this asymmetry in transition rates.