Although microbial ecology of the gut is now a major focus of interest, little is known about the molecular determinants of microbial adaptation in the gut. Experimental evolution coupled with whole-genome sequencing can provide insights of the adaptive process. In vitro experiments have revealed some conserved patterns: intermediate convergence, and epistatic interactions between beneficial mutations and mutations in global regulators. To test the relevance of these patterns and to identify the selective pressures acting in vivo, we have performed a long-term adaptation of an E. coli natural isolate, the streptomycin-resistant strain 536, in the digestive tract of streptomycin-treated mice. After a year of evolution, a clone from 15 replicates was sequenced. Consistently with in vitro observations, the identified mutations revealed a strong pattern of convergence at the mutation, gene, operon and functional levels. Yet, the rate of molecular evolution was lower than in in vitro, and no mutations in global regulators were recovered. More specific targets were observed: the dgo operon, involved in the galactonate pathway that improved growth on D-galactonate, and rluD and gidB, implicated in the maturation of the ribosomes, which mutations improved growth only in the presence of streptomycin. As in vitro, the nonrandom associations of mutations within the same pathways suggested a role of epistasis in shaping the adaptive landscape. Overall, we show that 'evolve and sequence' approach coupled with an analysis of convergence, when applied to a natural isolate, can be used to study adaptation in vivo and uncover the specific selective pressures of that environment.