one must realize that such simulations imply an enormous spread of length and time scales: even the simple case of a linear polymer exhibits geometrical structures from the scale of a chemical bond to the scale of the gyration radius, to the collective length scales that might be much larger in dense materials. [ 1 ] Moreover, our drive for modelling is not limited to linear polymers but also is directed at tree-like branched structures, cross-linked polymers, and even infinite molecular networks, i.e., gels. In this paper we propose a strategy that draws a consensus between what can be computed on the one hand and pursuing fundamental roots of the matter at hand on the other.The process of assembly or polymerization of molecular networks usually goes through a few temporal stages: disconnected monomers, linear polymer, branched polymer, and gel (an infi nite network). Furthermore, the gel itself should not be perceived as a single, fi nal state of polymer topology. After the gel point the connectivity patterns of the network continue to evolve. [2][3][4] In fact, in many cases a major part of the whole polymerization process may occur in the gel regime. [ 5 ] In view of extremely long time scales associated with polymerization, namely, the time scales that might well exceed hours for some industrial processes, [ 6 ] or even years (drying of oil paint [ 7 ] ), direct atomistic simulation Formation of a molecular network from multifunctional precursors is modelled with a random graph process. The process does not account for spatial positions of the monomers explicitly, yet the Euclidean distances between the monomers are derived from the topological information by applying self-avoiding random walks. This allows favoring reactivity of monomers that are close to each other, and to disfavor the reactivity for monomers obscured by the surrounding. As a result, the model is applicable to large time scales. The phenomena of conversion-dependent reaction rates, gelation, microgelation, and structural inhomogeneity are predicted by the model. Resulting nonhomogeneous network topologies are analyzed to extract such descriptors as: size distribution, crosslink distances, and gel-point conversion. Furthermore, new to the molecular simulation community descriptors are suggested that are especially useful when explaining evolution of the gel as being a single molecule: local clustering coeffi cient, network modularity, cluster size distribution.