Progress in retroviral gene transfer to large animal hematopoietic stem cells (HSCs) has led to efficient, reproducible long-term marking in both canine and nonhuman primate models. Successes for HSC gene therapy have occurred in the severe combined immunodeficiency setting, in which transduced cells have a selective advantage. However, for most diseases, the therapeutic transgene does not confer a sufficient survival advantage, and increasing the percentage of gene-marked cells in vivo will be necessary to observe a therapeutic effect. In vivo amplification should expand the potential of HSC gene therapy, and progress in this area has benefited greatly from the use of large animal models where efficacy and toxicity have often not correlated with results in murine models. To date, the best results have been observed with O(6)-methylguanine-DNA methyltransferase (MGMT) selection, with which increases in gene-marked repopulating cells have been maintained long-term, likely because of the toxicity of 1,3-bis-(2-chloroethyl)-1-nitrosourea and temozolomide to quiescent HSCs. Using MGMT selection, long-term marking levels exceeding 50% can now be routinely attained with minimal toxicity. There is cause to be optimistic that HSC gene therapy with in vivo amplification will soon allow the treatment of several genetic and infectious diseases.