Lanthanide-binding tag (LBT) peptides selectively complex lanthanide cations (Ln³⁺) in their binding pockets and are promising for lanthanide separation. However, designing LBTs that selectively target specific Ln³⁺ cations remains a challenge due to limited molecular-level understanding and control of interactions within the lanthanide-binding pocket. In this study, we reveal that the N5 asparagine residue acts as a gatekeeper in the binding pocket, resulting in a 100-fold selectivity for smaller Lu³⁺ over larger La³⁺ cations. Nuclear magnetic resonance spectroscopy and molecular dynamics simulations show that the N5 residue weakly binds to the larger La³⁺ cation, permitting H₂O molecules inside the pocket. For the smaller Lu³⁺ cations, the N5 residue forms an inter-arm hydrogen bond with the E14 glutamic acid residue, locking the Lu³⁺ cation in the pocket and preventing H₂O infiltration. Mutating the N5 asparagine to a D5 aspartic acid prevents such a hydrogen bond, eliminating the gatekeeping mechanism and precipitously reducing selectivity. The resulting binding affinity to Ln³⁺ cations is non-monotonic but generally increases with cation size. These results suggest a molecular design paradigm: the reduced affinity for larger lanthanides is due to open pocket conformations, while the selectivity of smaller Ln³⁺ cations over larger ones is due to the gatekeeping hydrogen bond.