The RNA World model represents a suitable scenario for the transition from prebiotic chemistry to DNA-based cellular life. Thus, it is currently assumed that Darwinian evolution originated when the first template-dependent RNA polymerase ribozyme emerged and genetic information was –unfaithfully– replicated for the first time. Nevertheless, a main unsolved problem is how an RNA polymerase ribozyme could have emerged from short RNA oligomers obtained by random polymerization on mineral surfaces. Indeed, a minimum size of 165 nt long is assumed for such a ribozyme, a length three to four times that of the longest RNA oligomers obtained by random polymerization on mineral surfaces. We have studied the structural repertoire yielded by large pools –up to 10E8 sequences– of random, 12 to 40 nt-long, ssRNA molecules. Our analysis shows that topologically simple RNA modules are the most abundant ones, especially hairpin structures and stem-loops (Stich et al., J. Theor. Biol. 2008). A fraction of the ubiquitous hairpin modules could have displayed RNA ligase activity, as they do in current ribozymes, and catalyze the assembly of larger, eventually functional, multi-modular RNA molecules. Therefore, structural and functional complexity could progressively increase even in the absence of template replication (Manrubia and Briones, RNA 2007). This allows us to build up a stepwise model of ligation-based, modular evolution that shows two main advantages with respect to previous hypotheses put forward for the origin of the RNA world: i) short RNA modules resulting from template-independent polymerization on different microenvironments might suffice to produce the first functional RNAs; ii) modular evolution shortens adaptation times and allows attaining complex structures that could not be otherwise directly selected (Briones et al., RNA 2009). Therefore, modular evolution could pave the way to the emergence of a ribozyme with RNA replicase activity at the dawn of the information-driven, Darwinian evolution.