In the age of "omics", lipidomics of erythropoiesis is still missing. How reticulocytes mature in the circulation into functional erythrocytes is also largely unknown. We have isolated here two populations of human circulating reticulocytes at different levels of maturation, and three subpopulations of erythrocytes of different age, and characterized the evolution of their lipidome. (Sphingomyelin+cholesterol) and partly phosphatidylethanolamine increase relative to total lipids, whereas phosphatidylcholine and phosphatidylserine decrease from immature reticulocytes to mature erythrocytes, at the same time as the surface area per cell decreases. The relative amounts of more than 70 phospholipid subclasses, based on the number of carbon atoms (12-24) and of double bonds (0-6) in the fatty acids linked to the phospholipid, also change in the process. As reticulocytes and erythrocytes cannot perform de-novo phospholipid synthesis, lipid remodeling likely requires selective removal of phospholipids from the membrane or their exchange with plasma or both, with the possible involvement of lipid transfer proteins such as VPS13A, which is expressed in reticulocytes and erythrocytes. These findings not only shed light on fundamental aspects of red blood cell physiology and erythropoiesis but also raise intriguing questions surrounding protein-lipid interactions, membrane architecture, and lipid trafficking mechanisms.