The success of allogeneic hematopoietic cell transplantation (allo-HCT) in curing hematological malignancies relies on the ability of donor-derived immune cells to recognize and eliminate residual tumor cells. However, in a far-from-negligible proportion of patients, after a first phase of remission cancer cells reappear (relapse). The historical explanation for occurrence of relapse was that treatment were not potent enough, leading as a consequence to increase doses and toxicities, often without significant benefit. A more modern perspective on relapse, to which our group has significantly contributed, is that tumor are smart moving targets, and that under the selective pressure of treatments they can modify their genetic and epigenetic features to select resistant variants (evolutionary model). We provided relevant proof-of-principles for this model, and showed that relapse can occur through a number of genomic (the genetic loss of HLA molecules) and non-genomic mechanisms (the epigenetically-driven modulation of HLA molecules and immune checkpoints). In addition, further evidence shows how, leukemic cells can also accelerate the progression of the disease by modifying the bone marrow microenvironment to their advantage, influencing the niche and exploiting immunoregulatory cell subsets.