Regression equations for the estimation of stature from long bones, although derived from modern human populations, are frequently applied to early hominids. In fact, some of these equations have even been recommended or especially created to be applied to Australopithecus remains. In this study, 45 sets of regression and correlation formulae, recurrent in anthropological and medico-legal literature, are applied to long bones of the Pliocene hominid A.L.288-1 (‘Lucy’), in order to assess which, if any, could be considered suitable for stature reconstruction in ‘gracile’ australopithecines. Virtually every method based on regression equations overestimates stature as compared with the estimate based on reconstruction of all the preserved skeletal parts. In addition, most methods failed to give consistent results with data from different limb segments. None of the sets of regression formulae tested here can be recommended as a reliable means of stature estimation in ‘gracile’ australopithecines.