The direct measurement of the incidence of new infections with the human immunodeficiency virus type 1 (HIV-1) can be made among soldiers because of the routine and periodic nature of HIV-1 testing in the United States Army. Between November 1985 and October 1993, 978 HIV-1 seroconversions were seen among 1,061,768 soldiers, contributing over 3.6 million person-years of follow-up [seroconversion rate (95% confidence interval) = 0.27/1,000 person-years (0.25-0.29)]. A significant decreasing trend in HIV-1 seroconversion rates was seen over the analysis period. The rate of new infections declined significantly from the first interval, 1985-1987, (0.43/1,000 person-years) to the second interval, 1987-1988, (0.28/1,000 person-years), but stabilized at approximately 0.22/1,000 person-years after 1988, representing new infections in approximately 100-150 soldiers annually. The risk of seroconversion among active duty soldiers was significantly associated with racial/ethnic group, age, gender, and marital status. Surveillance of HIV-1 seroconversion rates in the U.S. Army continues to offer a unique opportunity to assess temporal trends in the evolving HIV-1 infection epidemic. Monitoring the rate of new HIV-1 infections allows for identification of subgroups in need of intervention, refocusing of intervention strategies, and evaluation of their effectiveness.