Questions exist about whether testing of preventive human immunodeficiency virus (HIV)-1 vaccines, which will require rapid recruitment and retention of cohorts with high HIV-1 seroincidence, is feasible in the United States. A prospective cohort study was conducted in 1995-1997 among 4,892 persons at high risk for HIV infection in nine US cities. At 18 months, with an 88% retention rate, 90 incident HIV-1 infections were observed (1.31/100 person-years (PY), 95% confidence interval (CI): 1.06, 1.61). HIV-1 seroincidence rates varied significantly by baseline eligibility criteria--1.55/100 PY among men who had sex with men, 0.38/100 PY among male intravenous drug users, 1.24/100 PY among female intravenous drug users, and 1.13/100 PY among women at heterosexual risk-and by enrollment site, from 0.48/100 PY to 2.18/100 PY. HIV-1 incidence was highest among those men who had sex with men who reported unprotected anal intercourse (2.01/100 PY, 95% CI: 1.54, 2.63), participants who were definitely willing to enroll in an HIV vaccine trial (1.96/100 PY, 95% CI: 1.41, 2.73), and women who used crack cocaine (1.62/100 PY, 95% CI: 0.92, 2.85). Therefore, cohorts with HIV-1 seroincidence rates appropriate for HIV-1 vaccine trials can be recruited, enrolled, and retained.