Public health is paying increasing attention to elusive urban populations such as the homeless, street drug users, and illegal immigrants. Yet, valid data on the health of these populations remain scarce; longitudinal research, in particular, has been hampered by poor follow-up rates. This paper reports on the follow-up methods used in two randomized clinical trials among one such population, namely, homeless men with mental illness. Each of the two trials achieved virtually complete follow-up over 18 months. The authors describe the ethnographic approach to follow-up used in these trials and elaborate its application to four components of the follow-up: training interviewers, tracking participants, administering the research office, and conducting assessments. The ethnographic follow-up method is adaptable to other studies and other settings, and may provide a replicable model for achieving high follow-up rates in urban epidemiologic studies.