Adhesion of cells to solid surfaces plays an important role in many biological phenomena. Glass slides, coated with fibrinogen, have been covered with red blood cells (RBC) previously fixed with glutaraldehyde. The RBC solution is injected in a diffusion cell where the RBC's settle on the horizontal glass surface, which is observed with an inverted optical microscope (magnification 32x10). No rinsing is performed once all the RBC's are deposited. This experiment is repeated for a series of cell concentrations in the solution, ranging from 315 to 3740 cells/mm 3 . About hundred pictures are taken for each concentration. The relative surface covered by the RBC's, as well as the variance (or the standard deviation) of this surface over the sample of pictures available, is obtained by means of image processing.In order to model the process, the RBC's are approximated by monodisperse disks. A simulation code was developed on the basis of the R andom Sequential Adsorption (RSA) model, where the objects are deposited sequentially, at randomly chosen positions on the surface. Usually, no overlap is allowed, and the particles, once adsorbed, are supposed to stay permanently fixed in place. Here, however, the model is extended to account for possible overlaps between the RBC's. In fact, the surface covered, estimated by simulation, as a function of the number of particles deposited on the surface, agrees best with its measured counterpart when no overlap is assumed. Indeed, the observation at the microscope does not reveal many mutually overlaping cells, even at the highest concentration studied. In contrast, the RSA cannot account for the experimental coverage fluctuations. If the RBC's are treated as disks consisting of a hard core surrounded by a soft annulus, the variance of the surface covered can actually be raised by increasing the relative width of this annulus. However, the hypothesis of such soft particles is in contradiction with the already discussed linear relation between the surface covered and the number of particles deposited. This contradictory situation casts doubt on the validity of the RSA model for describing the RBC adhesion process. Instead, the ballistic deposition (BD) model seems to provide a better basis to describe the experimental observations. As far as the relation between the number of RBC's and the surface is concerned, it gives the same results as the RSA, provided that no overlap is allowed. As to the variance, it leads to a much larger one, though not in agreement with the measured variance over the whole range of concentrations investigated. In order to further improve the approch, the distribution of the cell radius (polydispersity) will be introduced in the model. Finally, whether the hydrodynamic interactions play a significant role in the observed configurations remains an open question.