During the last decade, small peptides (10 to 15 amino acids) derived from the HIV-1 Tat protein and from the drosophila Antennapedia homeodomain have been used to internalize various types of molecules into the cells. The way these peptides enter cells is still under investigation and the object of strong controversy. The main discussions rely on whether these peptides are internalized or not in an energy-independent fashion, and, depending on the situation, whether they follow one pathway instead of another. At present, we find in the literature a very large number of data with, at times, some contradictory results. Indeed the diversity of employed peptide sequences, the cell type used, the attachment or not of a cargo molecule, the chemical nature of this cargo itself, and the followed protocol during the experimental process do not simplify the comparison and hence final conclusions about the mechanism of cell entry. However, one common feature emerges with these cell-penetrating peptides: most of them do not show any cell specificity. Despite their demonstrated efficiency in delivering biologically active molecules in in vitro experiments, their use for a therapeutic application in vivo has been the object of a relatively little number of studies, probably because of the quite important amounts of CPP-cargo that needs to be prepared for an accurate and complete in vivo study, but more likely, because of the massive spreading of the cargo all around the body. However, it appears from recent studies that an increased targeting ability of these CPPs is possible, making the use of CPP mediated delivery compatible with an in vivo therapeutic approach.