Relating cellulose structure to its water uptake is a classical problem with many investigations done through measurements of cellulose "crystallinity". However, there is presently a growing consensus that crystallinity measurements are appreciably uncertain, leading to ambiguous interpretations of underlying cellulose organization. In this scenario, this article revisits the relations between cellulose structure and water uptake, moving the emphasis away from degree of crystallinity and directing it toward crystallite width, which is inferred with less ambiguity from the broadening of 200 X-ray diffraction peaks. With this approach, analysis of a wide spectrum of celluloses isolated from plants (preserving cellulose I phase and having variable contents of residual hemicelluloses) reveals a simple linear relation (R2 = 0.98) between reciprocal crystallite width and monolayer hydration (determined from vapor sorption). The primary role of crystallite width supports that most water-accessible polysaccharides are laterally associated with the crystallites, with a minor fraction in disordered domains along the fibrils. Furthermore, the secondary role left to hemicellulosic contents indicates cellulose being partly decrystallized to complement the disordered amount required to interface the crystallites. Finally, a substantial part of hydration is attributed to polysaccharides in voids left by the imperfect packing of aggregated crystallites.