In a multicenter, prospective, randomized, open-label study of isoniazid-preventive therapy (IPT) for latent tuberculosis infection, illegal immigrants from countries where tuberculosis is highly endemic were enrolled at two clinical sites in Northern Italy. Of 208 eligible subjects, 82 received supervised IPT at a dose of 900 mg twice weekly for 6 mo (Regimen A), 73 received unsupervised IPT 900 mg twice weekly for 6 mo (Regimen B), and 53 received unsupervised IPT 300 mg daily for 6 mo (Regimen C). Supervised IPT was delivered at either one tuberculosis clinic or one migrant clinic. The probability of completing a 26-wk regimen was 7, 26, and 41% in Regimens A, B, and C, respectively (p < 0.005, Log- rank test calculated using Kaplan-Meier plots). The mean time to dropout was 3. 8, 6, and 6.2 wk in Regimens A, B, and C, respectively (p = 0.003 for regimen A versus either Regimens B or C). Treatment was stopped in five subjects (2.4%) because of adverse events. The rate of completion of preventive therapy for latent tuberculosis infection among illegal immigrants was low. Supervised, clinic-based administration of IPT significantly reduced adherence. Alternative strategies to implement preventive therapy in illegal immigrants are clearly required.