The specific recognition of peptide sequences by proteins plays an important role both in biology and in diagnostic applications. Here we characterize the relatively weak binding of the peptide neurotensin (NT) to the previously developed Armadillo repeat protein VG_328 by a multidisciplinary approach based on solution NMR spectroscopy, mutational studies, and molecular dynamics (MD) simulations, totaling 20μs for all MD runs. We describe assignment challenges arising from the repetitive nature of the protein sequence, and we present novel approaches to address them. Partial assignments obtained for VG_328 in combination with chemical shift perturbations allowed us to identify the repeats not involved in binding. Their subsequent elimination resulted in a reduced-size binder with very similar affinity for NT, for which near-complete backbone assignments were achieved. A binding mode suggested by automatic docking and further validated by explicit solvent MD simulations is consistent with paramagnetic relaxation enhancement data collected using spin-labeled NT. Favorable intermolecular interactions are observed in the MD simulations for the residues that were previously shown to contribute to binding in an Ala scan of NT. We further characterized the role of residues within the N-cap for protein stability and peptide binding. Our multidisciplinary approach demonstrates that an initial low-resolution picture for a low-micromolar-peptide binder can be refined through the combination of NMR, protein design, docking, and MD simulations to establish its binding mode, even in the absence of crystallographic data, thereby providing valuable information for further design.