In the literature several competing hypotheses for the evolutionary mechanisms that shape metabolic pathways and the architecture of metabolic networks have been discussed, each of which finds support from comparative analysis of genomes. Alternatively, direct simulation studies on the the principles of metabolic evolution are rare because of the demanding pre-requisites. A central component of such a computational model is an algebraic chemistry model which acts as a substrate on which a metabolism can be selected. This component must be sufficiently involved to mimic the complexity of a modern metabolic network, without restricting the possible chemistry to the "known" extant end results. Furthermore, a genetic system that expresses catalysts and a non-trivial map from sequence/structure features of the catalysts to their respective functions within the metabolic network is necessary. The sequence to function map itself must be evolvable to allow the system to escape constraints set by the initial conditions. Finally, a fitness function that can be selected for and which evaluates metabolic efficiency is crucial. I will give a brief overview of the dominating mechanisms that governing pathway evolution and the architecture of modern metabolism, followed by an in-depth discuss of the various components that comprise our simulation framework. Finally results from large-scale evolutionary simulations will be presented. Literature: Ullrich A, Rohrschneider M, Scheuermann G, Stadler PF, Flamm C: In Silico Evolution of Early Metabolism. Artificial Life 17(2):87-108 (2011). Flamm C, Ullrich A, Ekker H, Mann M, Hoegerl D, Rohrschneider M, Sauer S, Scheuermann G, Klemm K, Hofacker IL, Stadler PF: Evolution of Metabolic Networks: A Computational Framework. J Sys Chem 1:4 (2010). Benkoe G, Flamm C, Stadler PF: A graph-based toy model of chemistry. J Chem Inf Comp Sci 43:1085-1093 (2003).