Genomic DNA is used as the template for both replication and transcription, whose machineries may collide and result in mutagenesis, among other damages. Because head-on collisions are more deleterious than codirectional collisions, genes should be preferentially encoded on the leading strand to avoid head-on collisions, as is observed in most bacterial genomes examined. However, why are there still lagging strand encoded genes? Paul et al. recently proposed that these genes take advantage of the increased mutagenesis resulting from head-on collisions and are thus adaptively encoded on the lagging strand. We show that the evidence they provided is invalid and that the existence of lagging strand encoded genes is explainable by a balance between deleterious mutations that bring genes from the leading to the lagging strand and purifying selection purging such mutants. Therefore, the adaptive hypothesis is neither theoretically needed nor empirically supported.