Stem cell gene therapy has long been limited by low gene transfer efficiency to hematopoietic stem cells. Recent years have witnessed clinical success in select diseases such as X-linked severe combined immunodeficiency (SCID) and ADA deficiency. Arguably, the single most important factor responsible for the increased efficacy of these recent protocols is the fact that the genetic correction provided a selective in vivo survival advantage. Since, for most diseases, there will be no selective advantage of gene-corrected cells, there has been a significant effort to arm vectors with a survival advantage. Two-gene vectors can be used to introduce the therapeutic gene and a selectable marker gene. Efficient in vivo selection strategies have been demonstrated in clinically relevant large-animal models. Mutant forms of the DNA repair-enzyme methylguanine methyltransferase in particular have allowed for efficient in vivo selection and have achieved sustained marking with virtually 100% gene-modified cells in large animals, and with clinically acceptable toxicity. Translation of these strategies to the clinical setting is imminent. Here, we review how in vivo selection strategies can be used to make stem cell gene therapy applicable to the treatment of a wider scope of genetic diseases and patients.