Osmotic dehydration (OD) is a coupled mass transfer operation where foods are partially dewatered while simultaneously impregnated with solute from the solution. Nonetheless, water loss ( W L ) and solute gain ( S G ) are treated as independent processes for modeling purposes and the effect of their coupling on the estimation of water/solute diffusivities ( D w , D s ) and equilibrium dehydration/impregnation levels ( W L e , S G e ) has not been elucidated yet. The objective of this study was to develop a new model considering the coupled mass transport of water and solute occurring during OD of foods and to evaluate the contribution of this coupling on the estimation of D w , D s , W L e , and S G e in comparison with the rigid/shrinking solid assumption. Experimental W L , S G , and volume kinetics were obtained during OD of papaya cubes in sucrose solutions (50°Bx; 45, 55 and 65 °C) and further analyzed with the proposed model solved under different assumptions: coupled and partially coupled mass transfer equations (MTEs) in a shrinking solid and uncoupled MTEs in a rigid solid (no shrinkage) with uncorrected and corrected lengths for diffusion. Properties D w , D s , W L e , and S G e with coupled MTEs and shrinkage were estimated in the ranges of (2.52–5.48) × 10 −10 m 2 /s, (3.91–5.43) × 10 −10 m 2 /s, 0.47–0.53 g/g and 0.18–0.30 g/g, respectively, with partially coupled and uncoupled MTEs leading to a significant overestimation of diffusivities under the tested experimental conditions (up to 281% for water and 75% for solute), even when shrinkage was considered in the 3D model solution. Current findings demonstrate that, like shrinkage, coupling of W L and S G is of upmost importance for a reliable estimation of mass transfer properties during OD. • A mass transfer model is proposed to analyze the OD of shrinking foods. • Mass transfer of water and solute are fully coupled by a density-related term. • OD experiments with papaya cubes were conducted to validate the model. • The effect of different modeling assumptions on mass transfer properties is analyzed. • Non-fully coupled models overestimate diffusivities even considering food shrinkage.