Epidemiological findings suggesting that formaldehyde exposure is associated with a higher risk of acute myelogenous leukemia (AML) and other hematological cancers have led to consideration of the potential mechanism of action by which inhalation of this rapidly reactive agent can cause bone marrow cancer. Two major mechanism-based arguments against formaldehyde as a leukemogen have been the difficulty in envisioning how inhaled formaldehyde might penetrate to the bone marrow; and the lack of similarity of non-cancer effects to other known human myeloleukemogens, particularly the absence of pancytopenia in humans or laboratory animals exposed to high levels. However, both of these arguments have been addressed by the recent finding of a pancytopenic effect and chromosomal abnormalities in heavily exposed Chinese workers which, if replicated, are indicative of a genotoxic effect of formaldehyde on hematopoietic stem cells that is in keeping with other known human leukemogens. Review of the body of evidence suggests an apparent discrepancy between studies in laboratory animals, which generally fail to show evidence of penetration of formaldehyde into the blood or evidence of blood or bone marrow genotoxicity, and studies of exposed humans in which there tends to be evidence of genotoxicity in circulating blood cells. One possible explanation for this discrepancy is species difference. Another possible explanation is that myeloid precursors within the nasal mucosa may be the site for leukemogenesis. However, chloromas, which are local collections of myeloid tumor cells, are rarely if ever found in the nose. Other proposed mechanisms for formaldehyde leukemogenesis are reviewed, and dose issues at the interface between the epidemiological and hematotoxicological findings are explored.