A large number of therapeutic drugs, herbal components and their metabolites are excreted by the kidneys. Therefore, generally applied models for estimating renal excretion, including freshly isolated rat proximal tubule cells, cultured tubule cells and immortalized kidney cell lines MDCKII, NRK-52E, IHKE-1 and Caki-1, were investigated regarding their predictive potential for active renal transport. Cultured proximal tubule cells showed an epithelial cell-like morphology and formed tight monolayers. However, mRNA expression analyses and immunohistochemical studies revealed patterns of tight junction proteins that were notably different from freshly isolated cells and distinct from those in vivo. High levels of mannitol permeation were found in NRK-52E, IHKE-1 and Caki-1 cells, suggesting that they are not suitable for bidirectional transport studies. Cultured cells and freshly isolated cells also differed in proximal tubule markers and transport proteins, indicating that cultured primary cells were in a state of dedifferentiation. Cell lines MDCKII, NRK-52E, IHKE-1 and Caki-1 did not accurately reflect the characteristics of proximal tubules. The expression patterns of marker and transport proteins differed from freshly isolated primary cells. In summary, each of these models has profound disadvantages to consider when adopting them reliable models for the in vivo situation. Thus, they should not be used alone but only in combination.