Marsupials show less diverse locomotor strategies than placentals. A common hypothesis for this discrepancy is that forelimb morphology is developmentally constrained in marsupial neonates due to the forelimb crawl from the birth canal to the pouch of the newborns. This would limit forelimb disparity and locomotor strategies more so in marsupials than in placentals. This developmental constraint hypothesis predicts that the less disparate forelimbs are less well correlated with the overall ecomorphological diversity than hindlimbs; forelimbs are less adapted to ecomorphologies than hindlimbs that are not involved in the neonatal crawl in marsupials. To test this prediction, we built a large dataset of limb skeletal metrics of major groups and all ecomorphotypes of marsupials, and we applied linear discriminant analyses and phylogenetic MANOVAs. Contrary to the predictions of the forelimb constraint hypothesis, didelphimorphian and diprotodontian forelimb morphology is often a better or equally good predictor of locomotor strategy than hindlimb morphology. Further, we show that this forelimb vs hindlimb difference can vary in additional marsupial groups (e.g., dasyuromorphians), in which hindlimbs are considerably better predictors of locomotor strategies as expected by developmental constraint hypothesis. These results suggest that developmental constraints on forelimbs, if present, are counter-balanced and can be overcome by selection associated with ecological traits in didelphimorphians and diprotodontians. On a macroevolutionary scale, the outcomes of the developmental constraint on forelimbs and ecological adaptation of both limbs can vary in different clades.