Specific protein-DNA interactions are central to a wide group of processes in the cell and have been studied both experimentally and computationally over the years. Despite the increasing collection of protein-DNA complexes, so far only a few studies have aimed at dissecting the structural characteristics of DNA binding among evolutionarily related proteins. Some questions that remain to be answered are: (a) what is the contribution of the different readout mechanisms in members of a given structural superfamily, (b) what is the degree of interface similarity among superfamily members and how this affects binding specificity, (c) how DNA-binding protein superfamilies distribute across taxa, and (d) is there a general or family-specific code for the recognition of DNA. We have recently developed a straightforward method to dissect the interface of protein-DNA complexes at the atomic level and here we apply it to study 175 proteins belonging to nine representative superfamilies. Our results indicate that evolutionarily unrelated DNA-binding domains broadly conserve specificity statistics, such as the ratio of indirect/direct readout and the frequency of atomic interactions, therefore supporting the existence of a set of recognition rules. It is also found that interface conservation follows trends that are superfamily-specific. Finally, this article identifies tendencies in the phylogenetic distribution of transcription factors, which might be related to the evolution of regulatory networks, and postulates that the modular nature of zinc finger proteins can explain its role in large genomes, as it allows for larger binding interfaces in a single protein molecule.