A theoretical analysis of experimental data is presented in this mini-review on non-random homo- and hetero-associations of cell surface receptors, which can be recruited in the plasma membrane or at the surface of the rough endoplasmic reticulum during the protein synthesis. In the latter case, the likely genetic origin of these supramolecular formations is analyzed, contrasting this concept to the mobility of the cell surface proteins. A model is offered which, on the one hand, allows the mobility in a restricted way even among microdomain-confined receptor proteins through 'swapping partners'. On the other hand, the lack of mixing molecular components of protein clusters will be analyzed, when homo-and hetero-associations are studied through cell fusion experiments. The most frequently studied cell surface patterns have included lipid raft organized HLA class I and II, ICAM-1, tetraspan molecules, IL2 and IL15 and other receptors, as well. On the contrary coated pit-associated transferrin receptors would not mix with the above lipid raft associated receptor patterns, although transferrin receptor would readily oligomerize into homo-associates. The functional consequences of these superstructures are also analyzed. On the 30th anniversary of the Singer-Nicolson fluid mosaic membrane model one has to pay tribute to the authors, because of their deep insight emphasizing also the mosaicism of the membranes in general and that of the plasma membrane, in particular.