Purpose: To investigate and compare the characteristics of four different types of retinal pigment epithelium (RPE) cells cultured for 2 to 5 weeks to provide guidance when choosing RPE cells for experimentation. Methods: Human cell lines ARPE-19 (ARPE) and D407, primary RPE cells from C57Bl/6 mouse (mRPE), and primary human fetal RPE (hfRPE) cells were grown in respective media previously reported to be optimal for each cell type. Two methods to obtain hfRPE were used: one isolated outside and transported to our laboratory, and one isolated primarily within our laboratory from donor human fetal eyes. Barrier function was determined by transepithelial electrical resistance (TER) and permeability and structure by localization of Na⁺,K⁺-ATPase α-1, ZO-1, and actin. VEGF expression, determined by real-time polymerase chain reaction (PCR) for mRNA and ELISA for protein, was determined after exposure to 24 h of 1% oxygen. Madin-Darby canine kidney (MDCK) cells were compared as a non-RPE epithelial cell line. Results: ARPE at passage 15, but not passage 32, maintained steady low TER measurements (up to 30 ohms*cm²) despite forming a monolayer with apical Na⁺,K⁺-ATPase α-1 labeling after 35 days. mRPE developed and maintained a TER of 30 ohms*cm² for 2 weeks but did not localize ATPase. hfRPE showed two phenotypes. hfRPE isolated remotely and sent to us appeared more mesenchymal and undifferentiated (hfRPE-U) and had a slow but steady increase in measured TER to approximately 25 ohms*cm², whereas hfRPE isolated from donor eyes in our laboratory showed well-differentiated monolayers (hfRPE-D) with TER measurements > 500 ohms*cm² within 1 month of culture. TER measurements reflected permeability determined by the measurement of paracellular movement of sodium fluorescein. All human RPE cell types showed expression of VEGF mRNA and protein, and expression was upregulated by hypoxia in hfRPE and D407, but not in ARPE, which had constitutively high expression. ARPE expressed high levels of VEGF protein in media and cell lysates (777.2; 54.4 pg/mg protein, respectively), whereas hfRPE and D407 produced significantly less (media: 5.7 [p = 0.001], 323.6 pg/mg protein [p = 0.01]; lysate: 0 [p < 0.001], 3.5 pg/mg protein [p < 0.001], respectively). Conclusions: Primary RPE cells and those from cell lines had different responses to medium-term culture or hypoxic stress. Primary isolation of hfRPE cells with careful control of culture conditions to assure adequate differentiation is recommended when using this cell as an example of a highly polarized epithelium. For disease, use of RPE cells that do not require long-term culture are more efficient and may be more relevant to study certain pathologies.