Risk classification and tailoring of treatment are essential for improving outcome for patients with acute myeloid leukemia or high-risk myelodysplastic syndrome. Both patient and leukemia-specific characteristics assessed using morphology, cytogenetics, molecular biology, and multicolor flow cytometry are relevant at diagnosis and during induction, consolidation, and maintenance phases of the treatment. In particular, minimal residual disease (MRD) during therapy has potential as a prognostic factor of outcome, determination of response to therapy, and direction of targeted therapy. MRD can be determined by cell surface markers using multicolor flow cytometry, whereas leukemia-specific translocations and mutations are measured using polymerase chain reaction–based techniques and recently using next-generation sequencing. All these methods of MRD detection have their (dis)advantages, and all need to be standardized, prospectively validated, and improved to be used for uniform clinical decision making and a potential surrogate end point for clinical trials testing novel treatment strategies. Important issues to be solved are time point of MRD measurement and threshold for MRD positivity. MRD is used for stem cell transplantation (SCT) selection in the large subgroup of patients with an intermediate risk profile. Patients who are MRD positive will benefit from allo-SCT. However, MRD-negative patients have a better chance of survival after SCT. Therefore, it is debated whether MRD-positive patients should be extensively treated to become MRD negative before SCT. Either way, accurate monitoring of potential residual or upcoming disease is mandatory. Tailoring therapy according to MRD monitoring may be the most successful way to provide appropriate specifically targeted, personalized treatment.