The safety of gene therapy using hematopoietic stem cells may be increased by including a suicide gene in the therapeutic vector to eliminate adverse events like insertional oncogenesis while retaining the clinical benefits. We have developed a model of experimental insertional oncogenesis by transducing the murine factor-dependent leukemia cell line Ba/F3 with a bicistronic Moloney murine leukemia virus retroviral vector encoding a murine oncogene (cKit(D814V)) in addition to one of three suicide genes: Herpes simplex virus thymidine kinase (HSV-TK); SR39, an HSV-TK mutant with an increased affinity for the drug substrate Ganciclovir (GCV); or sc39, a splice-corrected version of SR39. Following intravenous challenge with transduced Ba/F3 clones and treatment with GCV, leukemia developed in mice given cells expressing HSV-TK, but not SR39 or sc39. In vitro GCV resistance was observed in heterogeneously transduced Ba/F3 pools at 2.5-14%, and single-nucleotide changes or partial loss of the suicide gene were identified as mechanisms of drug escape. However, GCV treatment resulted in 80-100% survival of mice challenged even with pools of partially resistant Ba/F3 cells expressing SR39 or sc39. Thus, in this model of vector-driven insertional oncogenesis, a suicide gene approach was effective for eliminating leukemia using modified HSV-TK variants with improved biological activity.