Two in vitro methods were tested to establish their potential to predict the metabolizable energy (ME) content of forage legumes: the Tilley and Terry (TT) method and the pepsin-cellulase method (CM). Different samples of white clover (Trifolium repens L.), red clover (Trifolium pratense L.), kura clover (Trifolium ambiguum M. Bieb.), lucerne (Medicago sativa L.), and birdsfoot trefoil (Lotus corniculatus L.) were derived from field trials with several defoliation systems at two sites. The CM was more precise due to its repeatability within and between analysis runs, but eventually overestimated the ME contents of the samples, as it was shown for the standard samples with known in vivo digestibility. ME contents were found to be consistently higher based on CM, with a difference of up to 1.5 MJ ME/kg DM compared to TT. Although white clover was, in general, the species with the highest ME content, the influence of legume species over all cuts and defoliation systems was inconsistent. Such observations may influence the method of choice for ME estimation for large datasets.