Binding of all-trans Retinoic Acid (RA) to Cellular Retinoic Acid Binding Protein I (CRABP I) does not result in significant changes of the protein tertiary structure, even though the binding site is inaccessible in a static apo-protein conformation. One of the proposed scenarios for the protein-ligand binding process invokes the notion of a flexible portal region adjacent to the binding site, while another model suggests that the requisite dynamic events are induced by dimerization of the apo-protein in solution. In this work, RA binding to CRABP I is studied in dilute solutions (low micro-molar range), where no dimer and/or oligomer formation occurs. Modulation of backbone dynamics within various segments of the protein by its ligand is assessed using a combination of hydrogen exchange, electrospray ionization mass spectrometry, and collision-induced dissociation of protein ions in the gas phase. Consistent with the portal model of ligand entry, several protein segments (most of them containing residues making hydrophobic contacts to RA in the holo-form of the protein) are flexible in the absence of the ligand. At the same time, the two segments containing arginine residues forming a salt bridge with RA form the least flexible region in the apo-form of the protein. Although the presence of RA in solution reduces flexibility of all protein segments, the largest effect is observed within four strands that form one of the two beta-sheets enveloping a cavity which houses the ligand-binding site. These results are consistent with a model in which ligand binding occurs through a partially unstructured state of the protein with unobstructed access to the ligand-binding site. This intermediate (whose core is formed by the two stable arginine-containing strands) corresponds to a relatively low-energy local minimum on the apo-protein energy surface and is frequently sampled under native conditions.