The left panel shows a schematic representation of the numerous HDL particles from large to small spheres, discs, and poorly lipidated particles that one might encounter in a drop of peripheral blood. The problem with current subfractionation techniques is highlighted by the depicted ranges for methods that separate particles by size, density or charge. For example, isolated HDL3 particles are homogeneous with respect to density, but clearly contain particles of highly varying composition. Sizing methods and charge based methods suffer from the same issue. The yellow triangle shows how the generic marker of cholesterol not only encompasses all of the heterogeneity, but also overemphasizes lipid-rich, light, and large HDL particles while underemphasizing lipid-poor, heavy and small HDL particles including minimally lipidated apolipoproteins. The right panel shows the concept of compositionally defined subspecies containing active protein or lipid components that perform specific functions in different physiological processes. These species can be targeted by applying isolation or analytical techniques that target a specific component of the subspecies. We argue that it is these component-defined subspecies that need to be isolated and characterized if we are to make progress in understanding how HDL may be leveraged for disease therapies.