A cohort of 839 young girls at the ages of 14 and 15 years was screened for total antibodies to herpes simplex virus (HSV) and, if positive, for specific antibodies to HSV-2, by means of a sensitive, enzyme-linked immunosorbent assay (ELISA). The cohort was followed from 1972-1987. Blood samples were obtained on six occasions during these 16 years. In total, 2270 blood samples were taken. The number of sero-converting girls was studied in relation to calendar time. Two methods were constructed for the statistical analyses. The first of these gave an estimate of the sero-prevalence at different points in time. This analysis showed that the sero-prevalence which was 23% against HSV-1 in 1972 had increased to 36% in 1976. At the end of the study in 1987, 50% of the cohort had sero-converted against HSV-1. The proportion of girls who had sero-converted against HSV-2 was 0.4% in the 14-15-year-olds and had reached 22% by the end of the study. The second statistical method used all the available information implicit in the observations so as to obtain a maximum-likelihood (ML) estimate of the prevalence. The ML estimates were slightly more precise, but the two estimates did not differ significantly. The observations were further analysed by the Mantel-Haenszel test in order to see if there was any dependence between positivity to HSV-1 and HSV-2 respectively but none was found.