Two thirds of all Indian women have migrated for marriage, around 300 million women, but little is known about this vast migration. This paper provides a detailed accounting of the puzzlingly large migration of Indian women and evaluates its causes. Contrary to conventional wisdom, marriage migration does not contribute to risk sharing. Nor is it driven by sex ratio imbalances. Instead, I introduce a simple model in which parents must search for a spouse for their daughter geographically. By adding geographical search frictions, the model helps rationalize the large regional differences.