Numerous new fossils have driven an interest in reproduction of early birds but direct evidence remains elusive. No Mesozoic avian eggs can be unambiguously assigned to a species, which hampers our understanding of the evolution of contact incubation, which is a defining feature of extant birds. Compared to living species eggs of Mesozoic birds are relatively small, but whether the eggs of Mesozoic birds could actually have borne the weight of a breeding adult has not yet been investigated. We estimated maximal egg breadth for a range of Mesozoic avian taxa from the width of the pelvic canal defined by the pubic symphysis. Known elongation ratios of Mesozoic bird eggs allowed us to predict egg mass and hence the load mass an egg could endure before cracking. These values were compared to the predicted body masses of the adult birds based on skeletal remains. Based on 21 fossil species, we show that for non-ornithothoracine birds body mass was 130% of the load mass of the eggs. For Enantiornithes body mass and egg load mass were comparable to extant birds, but some early Cretaceous ornithuromorphs were 110% heavier than their eggs could support. Our indirect approach provides the best evidence yet that early birds could not have sat on their eggs without running the risk of causing damage. We suggest that contact incubation evolved comparatively late in birds.