In vivo selection strategies of genetically modified cells carrying mutant forms of MGMT (i.e. P140K or G156A) have the potential to improve autologous and allogeneic stem cell gene therapy and transplantation. Previously, we have shown efficient in vivo selection and marrow protection in a clinically relevant canine model. Here we describe molecular analysis of stem cell pools before, during, and after aggressive chemotherapy with O6BG and BCNU or temozolomide. LAM-PCR was carried out to identify clones and track those clones over time. This study includes four dogs that received chemotherapy; two dogs received DLA-identical allogeneic transplants (G154/G069) and two dogs received autologous transplants (G197/G179). A control dog received an autologous transplant, but never received chemotherapy (G038) was included. Chemotherapy treated animals received CD34-enriched marrow cells transduced with an RD114-pseudotype retroviral vector encoding P140K and eGFP while the control dog retroviral vector encoded only eGFP. After stable engraftment four dogs were treated with either O6BG and BCNU or temozolomide. Initially, granulocyte marking levels ranged from 10–16% before the dogs received repeated dose-escalating regimens of chemotherapy. After the final round of chemotherapy granulocyte marking was >98% and stabilized at 66–97% with stable increases in all cell lineages analyzed. For LAM-PCR analysis DNA was extracted from peripheral blood leukocytes before chemotherapy (71–95 days post-transplant designated PRE), after 4–6 treatments with chemotherapy (262–354 days post-transplant designated EARLY), and after the last treatment with chemotherapy (476–635 days post-transplant designated LATE). The DNA samples analyzed for G038 corresponded chronologically with that of the treatment animals with samples taken 95, 235, and 601 days post-transplant. Retroviral integration site analysis has shown that the four chemotherapy treated dogs and control dog are polyclonal at all times with no trend towards oligoclonal or monoclonal states. The stem cell pools of G038 are very similar with ~26% of clones identified in every pool after random sampling on bulk DNA. Additionally, overlapping sequences account for ~11% of PRE/EARLY clones, ~16% EARLY/LATE clones, and ~16% PRE/LATE clones. Unique sequences in each group account for about half of the clones analyzed. Conversely, in the dogs treated with chemotherapy, in over 40 clones analyzed no overlap between sequences has been identified. To further characterize the contribution of unique clones specific primers were designed to clones attained from integration sites collected in FACS-sorted populations from G197. When additional FACS-sorted and bulk peripheral blood samples from PRE/EARLY/LATE were probed the specific clone could be identified. This suggests that although the stem cell pool is polyclonal at all time points the clonal contribution of the pool is markedly different. This also suggests that the initial stem cell pool is composed of 100s to 1000s of clones. Retroviral copy number and telomere length of marked cells that have been selected by repeated treatments with chemotherapy is currently ongoing. Our studies suggest that even after aggressive in vivo selection of chemoprotected cells hematopoiesis has remained polyclonal and no side effects or malignancies were observed.