Genomes contain not only information for current biological functions, but also information for potential novel functions that may allow the host to adapt to new environments. The field of experimental evolution studies that potential by selecting for novel functions and deducing the means by which the function evolved, but until now it has not attempted to predict the outcomes of such experiments. Here I present a model system that is being developed specifically to examine the issue of what kind of information is most useful in predicting how novel functions will evolve. The system is the evolution of a Lac-PTS transport system and a phospho-beta-galactosidase hydrolase system as a novel pathway for metabolism of lactose in Escherichia coli. Two kinds of information, sequence-based phylogenetic inference and biochemical activity, are considered as predictors of which E. coli genes will evolve the required new functions. Both biochemical data and phylogenetic inference predict that the cryptic celABC genes, which currently specify a PTS-beta-glucoside transport system, are most likely to evolve into a PTS-lactose transport system. Phylogenetic inference predicts that the bglA gene, which currently specifies a phospho-beta-glucosidase, is most likely to evolve into a phospho-beta-galactosidase. In contrast, biochemical data predict that the cryptic bglB gene, which also currently specifies a phospho-beta-glucosidase, is most likely to evolve into a phospho-beta-galactosidase.