Advances in understanding the epidemiology of endometriosis have lagged behind other diseases because of methodologic problems related to disease definition and control selection. Nevertheless, a better picture of the epidemiology of endometriosis has emerged over the past few decades. Prevalence estimates of the disease in clinic populations vary from about a 4% occurrence of largely asymptomatic endometriosis found in women undergoing tubal ligation to 50% of teenagers with intractable dysmenorrhea. General population incidence during the 1970s in this country has been suggested to be 1.6 per 1000 white females aged 15-49, while a more current study based upon hospital discharges finds endometriosis as a first listed diagnosis in 1.3 per 1000 discharges in women aged 15-44. There is a clinical impression that blacks have lower rates of endometriosis and Orientals have higher rates than whites. A variety of personal risk factors for endometriosis have also been described. Women with endometriosis may be taller and thinner. Menstrual factors reported to increase risk include dysmenorrhea, early menarche, and shorter cycle lengths. There is support for the idea that lifestyle exposures that might raise or lower estrogen levels could affect risk, including a decreased risk associated with smoking and exercise and an increased risk associated with caffeine or alcohol use. These risk factors appear to be compatible with the central importance of retrograde menstruation influenced by outflow obstruction that might affect its amount, immune factors that might affect its ability to be cleared, or hormonal stimuli that might affect its growth. In this model, dysmenorrhea could be either a disease symptom or a manifestation of outflow obstruction. Nulliparity could be either a consequence of disease or a cause since nulliparous women would not have the benefit of cervical dilation associated with labor and delivery. Since there is evidence that family history is a risk factor for disease, a challenge is how to integrate genetic factors into the model. The genetics of endometriosis might be advanced if we could identify an "endometriosis phenotype". We propose that this may consist of early menarche, short cycles, painful periods, subfertility, and possibly tall stature that could be explained by genetic factors that predispose to poor endowment of germ cells and canalization defects of the cervix. The value of establishing an "endometriosis phenotype" is that, as candidates for genetic markers are identified, particular genotypes can be correlated with these factors even if a formal diagnosis of endometriosis has not been made. Additional well-designed case-control and cohort studies will be necessary to test theories related to pathogenesis, establish the precise relationship between reproductive morbidity and endometriosis, identify specific genetic factors, and establish long-term risks.