Across the water sector, Escherichia coli is the preferred microbial water quality indicator and current guidance upholds that it indicates recent faecal contamination. This has been challenged, however, by research demonstrating growth of E. coli in the environment. In this study, we used whole genome sequencing to investigate the links between E. coli and recent faecal contamination in drinking water. We sequenced 103 E. coli isolates sampled from 9 water supplies in rural Kitui County, Kenya, including points of collection (n = 14) and use (n = 30). Biomarkers for definitive source tracking remain elusive, so we analysed the phylogenetic grouping, multi-locus sequence types (MLSTs), allelic diversity, and virulence and antimicrobial resistance (AMR) genes of the isolates for insight into their likely source. Phylogroup B1, which is generally better adapted to water environments, is dominant in our samples (n = 69) and allelic diversity differences (z = 2.12, p = 0.03) suggest that naturalised populations may be particularly relevant at collection points with lower E. coli concentrations (<50 / 100mL). The strains that are more likely to have originated from human and/or recent faecal contamination (n = 50), were found at poorly protected collection points (4 sites) or at points of use (12 sites). We discuss the difficulty of interpreting health risk from E. coli grab samples, especially at household level, and our findings support the use of E. coli risk categories and encourage monitoring that accounts for sanitary conditions and temporal variability.