Understanding the interaction of graphene-related materials (GRM) with human cells is a key to the assessment of their potential risks for human health. There is a knowledge gap regarding the potential uptake of GRM by human intestinal cells after unintended ingestion. Therefore the aim of our study was to investigate the interaction of label-free graphene oxide (GO) with the intestinal cell line Caco-2 in vitro and to shed light on the influence of the cell phenotype given by the differentiation status on cellular uptake behaviour. Internalisation of two label-free GOs with different lateral size and thickness by undifferentiated and differentiated Caco-2 cells was analysed by scanning electron microscopy and transmission electron microscopy. Semi-quantification of cells associated with GRM was performed by flow cytometry. Undifferentiated Caco-2 cells showed significant amounts of cell-associated GRM, whereas differentiated Caco-2 cells exhibited low adhesion of GO sheets. Transmission electron microscopy analysis revealed internalisation of both applied GO (small and large) by undifferentiated Caco-2 cells. Even large GO sheets with lateral dimensions up to 10 µm, were found internalised by undifferentiated cells, presumably by macropinocytosis. In contrast, no GO uptake could be found for differentiated Caco-2 cells exhibiting an enterocyte-like morphology with apical brush border. Our results show that the internalisation of GO is highly dependent on the cell differentiation status of human intestinal cells. During differentiation Caco-2 cells undergo intense phenotypic changes which lead to a dramatic decrease in GRM internalisation. The results support the hypothesis that the cell surface topography of differentiated Caco-2 cells given by the brush border leads to low adhesion of GO sheets and sterical hindrance for material uptake. In addition, the mechanical properties of GRM, especially flexibility of the sheets, seem to be an important factor for internalisation of large GO sheets by epithelial cells. Our results highlight the importance of the choice of the in vitro model to enable better in vitro-in vivo translation.