Which sex should care for offspring is a fundamental question in evolution. Invertebrates, and insects in particular, show some of the most diverse kinds of parental care of all animals, but to date there has been no broad comparative study of the evolution of parental care in this group. Here, we test existing hypotheses of insect parental care evolution using a literature-compiled phylogeny of over 2000 species. To address substantial uncertainty in the insect phylogeny, we use a brute force approach based on multiple random resolutions of uncertain nodes. The main transitions were between no care (the probable ancestral state) and female care. Male care evolved exclusively from no care, supporting models where mating opportunity costs for caring males are reduced-for example, by caring for multiple broods-but rejecting the "enhanced fecundity" hypothesis that male care is favored because it allows females to avoid care costs. Biparental care largely arose by males joining caring females, and was more labile in Holometabola than in Hemimetabola. Insect care evolution most closely resembled amphibian care in general trajectory. Integrating these findings with the wealth of life history and ecological data in insects will allow testing of a rich vein of existing hypotheses.