Genomic imprinting refers to genes that are silenced when inherited via sperm or via egg. The silencing of genes conditional upon their parental origin requires an evolutionary explanation. The most widely accepted theory for the evolution of genomic imprinting—the kinship theory—argues that conflict between maternally inherited and paternally inherited genes over phenotypes with asymmetric effects on matrilineal and patrilineal kin results in self-imposed silencing of one of the copies. This theory has been applied to imprinting of genes expressed in the placenta, and infant brain determining the allocation of parental resources being the source of conflict parental promiscuity. However, there is growing evidence that imprinted genes are expressed in the postinfant brain where parental promiscuity per se is no longer a source of conflict. Here, we advance the kinship theory by developing an evolutionary model of genomic imprinting in adults, driven by intragenomic conflict over allocation to parental versus communal care. We consider the role of sex differences in dispersal and variance in reproductive success as sources of conflict. We predict that, in hominids and birds, parental care will be expressed by maternally inherited genes. In nonhominid mammals, we predict more diversity, with some mammals showing the same pattern and other showing the reverse. We use the model to interpret experimental data on imprinted genes in the house mouse: specifically, paternally expressed Peg1 and Peg3 genes, underlying maternal care, and maternally expressed Gnas and paternally expressed Gnasxl genes, underlying communal care. We also use the model to relate ancestral demography to contemporary imprinting disorders of adults, in humans and other taxa.