Recurring patterns of primary structure have been observed in enzymes that mediate sequential metabolic reactions in bacteria. The enzymes, muconolactone Delta-isomerase [(+)-4-hydroxy-4-carboxymethylisocrotonolactone Delta(2)-Delta(3)-isomerase, EC 5.3.3.4] and beta-ketoadipate enol-lactone hydrolase [4-carboxymethylbut-3-enolide(1,4)enol-lactone-hydrolase, EC 3.1.1.24], have been coselected in bacterial populations because the isomerase can confer no nutritional advantage in the absence of the hydrolase. Similar amino acid sequences recur within the structure of the isomerase, and the amino-terminal amino acid sequence of the isomerase from Pseudomonas putida appears to be evolutionarily homologous with the corresponding sequence of a beta-ketoadipate enol-lactone hydrolase from Acinetobacter calcoaceticus. One interpretation of the sequence repetitions is that they reflect tandem duplication mutations that took place early in the evolution of the proteins. According to this view, the mutations caused elongation of structural genes and the creation of duplicated genes as the metabolic pathways evolved. A review of the sequence data calls attention to a different hypothesis: repeated amino acid sequences were introduced in the course of the proteins' evolution by substitution of copies of DNA sequences into structural genes. Our observations are interpreted on the basis of a model proposing genetic exchange between misaligned DNA sequences. The model predicts that misalignments in one chromosomal region can influence the nature of mutations in another region. Thus, as often has been observed, the mutability of a base pair will be determined by its location in a DNA sequence. Furthermore, the intrachromosomal recombination of DNA sequences may account for complex genetic modifications that occur as new pathways evolve. The model provides an interpretation of an apparent paradox, the rapid creation of new metabolic traits by bacterial genomes that are remarkably resistant to genetic drift.