We model the immune dynamics between T cells and cancer cells in leukemia patients after a bone-marrow (or a stem-cell) transplant. We use a system of nine delay differential equations that incorporate time delays and account for the progression of cells through different stages. This model is an extension of our earlier model [DeConde et. al, JTB, 236, 2005, 39-59]. We conduct a sensitivity analysis of the model parameters with respect to the minimum cancer concentration attained during the first remission and the time until the first relapse. In addition, we examine the effects of varying the initial host cell concentration and the cancer cell concentration on the likelihood of a successful transplant. We observe that higher initial concentrations of general host blood cells increase the chance of success. Such higher initial concentrations can be obtained, e.g., by reducing the amount of chemotherapy that is administered prior to the transplant, a procedure known as a mini-transplant. Our results suggest that mini-transplants may be advantageous over full transplants. We identify the regions of the parameters for which mini-transplants are advantageous using statistical tools.