Boron neutron capture therapy (BNCT) requires pharmaceutical innovations and molecular-based evidence of effectiveness to become a standard cancer therapeutic in the future. Recently, in Japan, 4-borono-L-phenylalanine (BPA) was approved as a boron agent for BNCT against head and neck (H&N) cancers. H&N cancer appears to be a suitable target for BPA-BNCT, because the expression levels of L-type amino acid transporter 1 (LAT1), one of the amino acid transporters responsible for BPA uptake, are elevated in most cases of H&N cancer. However, in other types of cancer including malignant brain tumors, LAT1 is not always highly expressed. To expand the possibility of BNCT for these cases, we previously developed poly-arginine peptide (polyR)-conjugated mercaptoundecahydrododecaborate (BSH). PolyR confers the cell membrane permeability and tumor selectivity of BSH. However, the molecular determinants for the properties are not fully understood. In this present study, we have identified the cluster of differentiation 44 (CD44) protein and translational machinery proteins as a major cell surface target and intracellular targets of BSH-polyR, respectively. CD44, also known as a stem cell-associated maker in various types of cancer, is required for the cellular uptake of polyR-conjugated molecules. We showed that BSH-polyR was predominantly delivered to a CD44High cell population of cancer cells. Once delivered, BSH-polyR interacted with the translational machinery components, including the initiation factors, termination factors, and poly(A)-biding protein (PABP). As a proof of principle, we performed BSH-polyR-based BNCT against glioma stem-like cells and revealed that BSH-polyR successfully induced BNCT-dependent cell death specifically in CD44High cells. Bioinformatics analysis indicated that BSH-polyR would be suitable for certain types of malignant tumors. Our results shed light on the biochemical properties of BSH-polyR, which may further contribute to the therapeutic optimization of BSH-BNCT in the future.