It is of crucial importance to modify dextran-based polysaccharides in the design of novel biomedical materials. A simple one-step method, involving the reaction of hydroxyl groups-of dextran with alpha-bromoisobutyric acid in the presence of 1, l'-carbonyldiimidazole, was first developed to produce bromoisobutyryl-terminated dextran as multifunctional initiators for subsequent atom transfer radical polymerization (ATRP). Well-defined comb-shaped copolymers (DPDs) composed of nonionic hydrophilic dextran backbones and cationic poly((2-dimethyl amino)ethyl methacrylate) (or P(DMAEMA)) side chains were subsequently prepared via ATRP for nonviral gene delivery. The P(DMAEMA) side chains of DPDs can be further partially quaternized to produce the quaternary ammonium DPDs (QDPDs) DPD and QDPDs can condense pDNA into complex nanoparticles of 100 to 150 nm in sizes. QDPDs exhibit stronger ability to complex pDNA, due to increased surface cationic charges. DPDs can exhibit much lower cytotoxicity and better gene transfection yield than high-molecular-weight P(DMAEMA) homopolymers and "gold-standard" polyethylenimine (25 kDa) in HEK293 and L929 cell lines. DPDs also exhibit efficient gene delivery ability-in different cancer cell lines, especially in MCF7 cells-where the DPD-mediated transfection efficiency is almost 3 times higher than that of the popular Lipfectamine 2000 transfection reagent. This study demonstrated that grafting low-molecular-weight polymer chains from natural dextran backbones via ATRP is an effective means to produce novel polysaccharide-based nanobiomaterials.