Caveolae, flask-shaped invaginations of the plasma membrane, have long been implicated in endocytosis. Recent data question this link, and in the absence of specific cargoes the potential cellular function of caveolar endocytosis remains unclear (1). Moreover, recent data showed that caveolar biogenesis requires precise ratio between caveolar proteins. Thus, perturbation in this ratio, such as by overexpression of caveolin-1, leads to its ubiquitination and a consequent degradation through classical endo-lysosomal pathway (2,3). This process affects caveolar dynamics and may be misinterpreted as caveolar endocytosis, thus hampering any attempts to study caveolar dynamics, as well as addressing the link between caveolae and endocytosis or intracellular membrane trafficking. To overcome these pitfalls we have utilized genome-edited techniques to label with fluorescent tag both caveolin-1 and cavin-1 proteins expressed at endogenous levels (4). These proteins, present at the “bulb” of this structure, essential for caveolae formation serve as known markers of caveola. We find that dynamics of endogenous caveolin-1-GFP or cavin-1-Cherry is very different from that of overexpressed proteins. We find that around 5% of the cellular pool of caveolae is present on dynamic endosomes, and is delivered to endosomes in a clathrin-independent manner. Furthermore, we show that caveolae are indeed likely to bud directly from the plasma membrane. Using a genetically encoded tag for electron microscopy and ratiometric light microscopy, we go on to show that bulk membrane proteins are depleted within caveolae. Although caveolae are likely to account for only a small proportion of total endocytosis, cells lacking caveolae show fundamentally altered patterns of membrane traffic when loaded with excess glycosphingolipid. Together, these observations support the hypothesis that caveolar endocytosis is specialised in maintenance of plasma membrane lipids.