OBJECTIVES--To examine the epidemiology of antibody to herpes simplex virus type 2 and to assess its suitability as a serological marker of sexual behaviour in populations with high and low prevalences. DESIGN--Cross sectional survey. SETTING--Department of genitourinary medicine and blood donation centre in central London. SUBJECTS--Representative sample of 869 patients attending department between November 1990 and December 1991, and 1494 consecutive blood donors attending for donation between February and April 1992. METHOD--Participants had a blood sample taken for antibody testing with a novel type specific assay and completed a questionnaire. RESULTS--Prevalence of antibody differed significantly between the two groups (188/833 (22.7%) clinic attenders; 102/1347 (7.6%) blood donors). In both populations antibody was strongly associated with sex, sexual orientation, years of sexual activity, number of lifetime sexual partners, and past infection with sexually transmitted diseases after other factors were controlled for. Only 130 (45%) of all those with antibody had symptoms suggestive of genital herpes, and 79 (27.4%) had had genital herpes diagnosed. Of those without antibody to herpes simplex viruses type 1 and 2, 8.0% reported genital blisters or sores and 1.1% had had genital herpes diagnosed by a doctor. CONCLUSIONS--The strong relation between herpes simplex virus type 2 and sexual lifestyle suggests that the presence of antibody to the virus may be suitable for use as an objective, serological marker of patterns of sexual behaviour in different populations. These data show that only a minority of those infected with herpes simplex virus type 2 have a diagnosis of genital herpes or express clinical symptoms, making serological determinants of infection essential for epidemiological studies.