Even after extended treatment with powerful antiretroviral drugs, HIV is not completely eliminated from infected individuals. Latently infected CD4(+) T cells constitute one reservoir of replication-competent HIV that needs to be eliminated to completely purge virus from antiretroviral drug-treated patients. However, a major limitation in the development of therapies to eliminate this latent reservoir is the lack of relevant in vivo models that can be used to test purging strategies. Here, we show that the humanized BLT (bone marrow-liver-thymus) mouse can be used as both an abundant source of primary latently infected cells for ex vivo latency analysis and also as an in vivo system for the study of latency. We demonstrate that over 2% of human cells recovered from the spleens of HIV-infected BLT mice can be latently infected and that this virus is integrated, activation inducible, and replication competent. The non-tumor-inducing phorbol esters prostratin and 12-deoxyphorbol-13-phenylacetate can each induce HIV ex vivo from these latently infected cells, indicating that this model can be used as a source of primary cells for testing latency activators. Finally, we show activation-inducible virus is still present following suppression of plasma viral loads to undetectable levels by using the antiretroviral drugs zidovudine, indinavir sulfate, and didanosine, demonstrating that this model can also be used to assess the in vivo efficacy of latency-purging strategies. Therefore, the HIV-infected BLT mouse should provide a useful model for assessment of HIV latency activators and approaches to eliminate persistent in vivo HIV reservoirs.