Sperm are vectors of disease. During mammalian insemination bacteria from the male and female genitalia regularly cling to sperm tails and are transported to the uterus. I propose that menstruation functions to protect the uterus and oviducts from colonization by pathogens. Menstrual blood exerts mechanical pressure on uterine tissue, forcing it to shed, and delivers large numbers of immune cells throughout the uterine cavity, directly combating pathogens. The mechanisms of menstruation show evidence of adaptive design. Spiral arteries that open to the lining of the uterus trigger menstruation by abruptly constricting, which deprives the local tissue of blood, and then abruptly dilating, which causes blood to force loose the necrotic tissue. Menstrual blood flows easily, unlike blood at most wound sites, because it lacks the normal level of clotting factors. Overt (externally visible) or covert (not externally visible) menstruation has been documented in many species of primate, including Old World monkeys and apes, New World monkeys, and prosimians, as well as in various species of bat and insectivore. The antipathogen hypothesis predicts that: (1) menstruation (overt or covert) is either universal or nearly so among mammalian species; (2) if the latter, then the existence of menstruation among species varies inversely with the probability of becoming pregnant per estrous cycle (menstruation would be especially adaptive in species with significantly less than 100% probability of becoming pregnant per estrous cycle); (3) among menstruating species, the average degree of menstrual bleeding for a given species is a function of the factors affecting menstruation's costs and benefits--in particular, the degree of bleeding is positively correlated with the average body size and sexually transmitted pathogen load of that species (profuse bleeding would be especially adaptive in large-bodied species with either promiscuous breeding systems or continuous sexual receptivity); and (4) other forms of normal uterine bleeding--proestrous, periovulatory, implantation, and postpartum--also have an antipathogen function. The hypothesis presented in this article has implications for the diagnosis, treatment and prevention of uterine infection and, therefore, for the prevention of pathogen-induced infertility. The uterus appears to be designed to increase its bleeding if it detects infection: Human uteri that become infected (or otherwise inflamed) bleed more profusely, bleed on more days per cycle, and often bleed intermittently throughout the cycle. Thus artificially curtailing infection-induced uterine bleeding may be contraindicated.