An unusual distribution of myeloperoxidase has been observed in HL-60 cells using enzyme cytochemistry and electron microscopy. Its localization in these cells differed in two ways from its previously observed distribution in normal human promyelocytes developing in vivo. First, the majority of Golgi cisternae and vesicles of HL-60 cells did not contain detectable peroxidase, whereas the rough endoplasmic reticulum (RER), granules, and dilated vacuoles were positive. This suggests that there is a block in vesicular transport from RER to the Golgi complex, or in fusion of the vesicles with Golgi membranes. Second, the peroxidase-positive granules that were formed in vitro varied greatly from those formed in vivo as follows: (1) their limiting membrane was narrower, (2) their contents were less concentrated, and (3) they resemble autophagic vacuoles rather than normal storage granules because they contained numerous membranes. These findings suggest that information gleaned from HL-60 cells about the biosynthetic pathway of myeloperoxidase and the formation of granules should be interpreted with caution.