Background: Whereas, in most patients with multiple myeloma (MM), achieving undetectable MRD anticipates a favorable outcome, some others relapse shortly afterwards. Although one obvious explanation for this inconsistency is the use of nonrepresentative marrow samples due to hemodilution, there is no guidance on how to evaluate this issue. Methods: Since B-cell precursors, mast cells and nucleated red blood cells are normally absent in peripheral blood, we analyzed them in 1404 bone marrow (BM) aspirates obtained in numerous disease settings and in 85 healthy adults (HA). Results: First, we confirmed the systematic detection of the three populations in HA, as well as the nonreduced numbers with aging. Pairwise comparisons between HA and MM patients grouped according to age and treatment showed significant variability, suggesting that hemodilution should be preferably evaluated with references obtained from patients treated with identical regimens. Leveraging the MRD results from 118 patients, we showed that a comparison with HA of similar age could also inform on potential hemodilution. Conclusions: Our study supports the routine assessment of BM cellularity to evaluate hemodilution, since reduced BM-specific cell types as compared to reference values (either treatment-specific or from HA if the former are unavailable) could indicate hemodilution and a false-negative MRD result.