During cheese ripening, bacteria grow as immobilized colonies, metabolizing substrates present in the matrix and generating products triggered by enzymatic reactions. Local limitation rates of diffusion, either in the matrix or around the colonies, may be responsible for modulation of the metabolic activities of bacteria. Therefore, diffusion properties of small solutes (e.g. bacterial metabolites) and macromolecules (e.g. nutrients and enzymes) may have a crucial role in the ripening process and in the final quality of cheese. Diffusion phenomena may depend both on cheese microstructure and on physicochemical characteristics of solutes. The determination of the relationship between diffusion properties of solutes and cheese microstructure is now an important step in order to better understand cheese ripening mechanisms. Floury et al. (2012) showed the great potentiality of Fluorescence Recovery after Photobleaching (FRAP), a sophisticated fluorescence-based technique, to study the diffusion of fluorescent solutes in cheese. We have used the FRAP technique to investigate the influence of different physicochemical characteristics of solutes (size, charge and shape) on their diffusion behavior in an ultrafiltered (UF)-model cheese. Moreover, there is a great interest in quantifying the various aspects of the model cheese structure. We have adapted several advanced microscopic techniques and innovative image processing methodologies to determine structural parameters in the model cheese. Quantification of structural features may allow us to relate diffusion phenomena to the microstructure of the UF-model cheese.[br/] [br/] Methodology Commercial FITC-dextrans of different molecular weights (4kDa to 2MDa) were chosen as models of linear/flexible solutes. FITC-dairy proteins (α-lactalbumin, β-lactoglobulin and bovine serum albumin) of different sizes were chosen for their rigid/globular shapes. Diffusion coefficients of fluorescent solutes were measured in an UFcheese using the FRAP technique. This model cheese consists in a fat-free dairy gel made from renneted UF-retentate. As described previously by Aly et al. (2011), this UF-model cheese has a great advantage over traditional cheese technology to be mold after renneting without further syneresis of the gel. Moreover, this matrix exhibits highly reproducible microand macro- structural properties. A large variety of microscopic techniques, including transmission electron microscopy (conventional TEM and cryo-TEM) and confocal laser scanning microscopy (CLSM), have been performed to study the microstructure of the model cheese. The goal is to find techniques that preserve the fine details of the UF-cheese structure. Moreover, digital image analysis has been developed to process images in order to extract numerical data referred to the microstructure. Results and discussion This study have showed that macromolecules up to 2 MDa and proteins could diffuse in the UF-model cheese. FITC-dextran diffusion data were fitted according to the obstruction model, resulting in a porosity/tortuosity ratio ~ 0.42. Diffusion in the model cheese was sensitive to the solute shape and charge. The FITC-dairy proteins studied (rigid and negatively charged) were more hindered than the FITC-dextrans (flexible and neutral) in the UF-model cheese. Computer image analysis of CLSM images has provided visualization of the pores and determination of the pore sizes of the UF-model cheese. Conventional TEM has been applied to determine the pore size, particle size and tortuosity of the matrix. Parallely, cryo-TEM has been used in order to minimize artifacts due to the sample preparation, preserving the native structure of the UF-model cheese. Indeed, different findings have been observed between images obtained from these two kinds of TEM techniques by using innovative digital image analysis. Image analysis of microscopy images in conjunction with modeling and simulation has been offering new possibilities for determining and understanding the structure of the UFmodel cheese. Moreover, tomographic 3D images have been reconstituted from TEM images. This technique represents a new promising approach to a full characterization of the internal structure of the UF-model cheese. Conclusions To be able to control mass transport in cheese, we need to understand the limitations of diffusion as a function of the microstructure, composition and heterogeneity of the product. Structure-based modeling has been used to determine the effect of the structure on the solute diffusion in a UF-model cheese using a novel combination of several microscopy techniques, 2D image analysis and 3D tomography. A better understanding of the influence of the matrix microstructure on the mobility of different solutes should allow a more generic view of cheese ripening kinetics for future innovations.