Transposable and interspersed repetitive elements (TIREs) are ubiquitous features of both prokaryotic and eukaryotic genomes. However, controversy has arisen as to whether these sequences represent useless 'selfish' DNA elements, with no cellular function, as opposed to useful genetic units. In this review, we selected two insect species, the Dipteran Drosophila and the Lepidopteran Bombyx mori (the silkmoth), in an attempt to resolve this debate. These two species were selected on the basis of the special interest that our laboratory has had over the years in Bombyx with its well known molecular and developmental biology, and the wealth of genetic data that exist for Drosophila. In addition, these two species represent contrasting repetitive element types and patterns of distribution. On one hand, Bombyx exhibits the short interspersion pattern in which Alu-like TIREs predominate while Drosophila possesses the long interspersion pattern in which retroviral-like TIREs are prevalent. In Bombyx, the main TIRE family is Bm-1 while the Drosophila group contains predominantly copia-like elements, non-LTR retroposons, bacterial-type retroposons and fold-back transposable elements sequences. Our analysis of the information revealed highly non-random patterns of both TIRE biology and evolution, more indicative of these sequences acting as genomic symbionts under cellular regulation rather than useless or selfish junk DNA. In addition, we extended our analysis of potential TIRE functionality to what is known from other eukaryotic systems. From this study, it became apparent that these DNA elements may have originated as innocuous or selfish sequences and then adopted functions. The mechanism for this conversion from non-functionality to specific roles is a process of coevolution between the repetitive element and other cellular DNA often times in close physical proximity. The resulting interdependence between repetitive elements and other cellular sequences restrict the number of evolutionarily successful mutational changes for a given function or cistron. This mutual limitation is what we call genome canalization. Well documented examples are discussed to support this hypothesis and a mechanistic model is presented for how such genomic canalization can occur. Also proposed are empirical studies which would support or invalidate aspects of this hypothesis.