Although anti-angiogenic therapy is predicted to be an effective therapy for treating cancer, selectively targeting tumor endothelial cells (TECs), and not normal endothelial cells, remains a major obstacle. Modifying a drug carrier with a targeting ligand is a popular strategy for developing an active-targeting type drug delivery system (DDS). We previously reported that a cyclo(Arg-Gly-Asp-d-Phe-Lys) (cRGD)-equipped liposome that contains encapsulated siRNA (RGD-MEND) achieved an efficient therapeutic outcome in a murine cancer model. To develop a more efficient TEC-targeting DDS, we examined the effect of the length of the polyethylene glycol (PEG) that is used as a peptide-linker on the cholesterol-scaffold, and liposomal composition on the efficiency of delivery of siRNA to cRGD receptor αVβ3 integrin positive cells. An RGD-MEND modified with shorter linker/no-linker, PEG350 or no-PEG, showed a higher cellular uptake in vitro. However, a shorter or no-linker RGD-cholesterol-modified MEND showed no silencing effect despite its high, in vitro silencing efficiency. To examine the possibility that the cholesterol-scaffold ligand was removed from the surface of the RGD-MEND by interactions with serum proteins, the RGD-MEND was incubated in the presence of a 50% serum solution. The cellular uptake of the cholesterol-scaffold ligand was drastically reduced by the incubation in serum. Increasing the cholesterol ratio in the lipid envelope and adding a helper lipid improved the in vivo knockdown efficiency, probably due to an enhanced ligand retention, even in in vivo conditions. The findings reported herein suggest that the lipid composition and the ligand scaffold of the MEND are major factors in successfully developing an efficient active-targeting DDS.