Understanding the mechanism of action of the antimicrobial peptide (AMP) in terms of its structure and energetics is the key to designing new potent and selective AMPs. Recently, we reported a membranolytic 14-residue-long lysine-rich cationic antimicrobial peptide (LL-14: NH3+-LKWLKKLLKWLKKL-CONH2) against Pseudomonas aeruginosa, Klebsiella pneumoniae, and Staphylococcus aureus, which is limited by cytotoxicity and expected to undergo facile protease degradation. Aliphatic side-chain-length modification of the cationic amino-acid residues (Lys and Arg) is a popular strategy for designing protease-resistant AMPs. However, the effect of the peptide side-chain length modifications on the membrane binding affinity and its relation to the atomic structure remain an unsolved problem. We report computer simulations that quantitatively calculated the difference in peptide binding affinity to membrane-mimetic-bilayer models (bacterial: 1-palmitoyl-2-oleoyl-phosphatidylethanolamine (POPE)/1-palmitoyl-2-oleoyl-phosphatidylglycerol (POPG) bilayer and mammalian: 1-palmitoyl-2-oleoyl-phosphatidylcholine (POPC) bilayer) upon decreasing or increasing the spacer length of the cationic lysine residues of LL-14 (as well as their arginine analogues). We show that the peptide/bilayer interaction energetics varies drastically in response to spacer length modification. The strength of peptide discrimination depends strongly on the nature of the bilayer (bacterial or mammalian mimetic model). An increase in the lysine spacer length by one carbon (i.e., homolysine analogue of LL-14) is weakly/strongly disfavored by the bacterial/mammalian-membrane-mimetic bilayer. Recently, we have demonstrated an excellent correlation between the antimicrobial activity of the membranolytic cationic peptides and their binding affinity to membrane-mimetic-bilayer models. Thus, the homolysine analogue of LL-14 is a promising noncytotoxic AMP with conserved activity. On the other hand, homoarginine analogue (arginine spacer length increment by a single carbon) was preferred by both the bacteria and the mammalian mimetic bilayers and displayed the strongest affinity for the former among the peptides studied in this work. Thus, the promising most potent homoarginine analogue is likely to be cytotoxic. Shortening the Lys/Arg side chain to a three-carbon spacer (Dab/Agb) improves the binding affinity to bacterial and mammalian-membrane-mimetic bilayers. Arginine and arginine-derivative peptides exhibited stronger binding affinity to the bilayers relative to the lysine analogue. The results provide a plausible explanation to the previous experimental observations, viz., superior antimicrobial activity of the arginine peptides relative to Lys peptides and the improvement of antimicrobial activity upon substitution of Lys with Dab in the cationic peptides. The simulations revealed that the small change in the peptide hydrophobicity by Lys/Arg spacer length modification could drastically alter the energetics of peptide/bilayer binding by fine-tuning the electrostatic interactions. The energetics underlying the peptide selectivity by simple membrane-mimetic bilayer models may be beneficial for designing new selective and protease-resistant AMPs.