Freshwater samples (n = 199) were obtained from 41 sites with contrasting land-uses (avian, low impact, dairy, urban, sheep and beef, and mixed sheep, beef and dairy) and the E. coli phylotype of 3980 isolates (20 per water sample enrichment) was determined. Eight phylotypes were identified with B1 (48.04%), B2 (14.87%) and A (14.79%) the most abundant. Escherichia marmotae (n = 22), and Escherichia ruysiae (n = 1), were rare (0.68%) suggesting that these environmental strains are unlikely to confound water quality assessments. Phylotypes A and B1 were overrepresented in dairy and urban sites (p < 0.0001), whilst B2 were overrepresented in low impact sites (p < 0.0001). Pathogens ((Salmonella, Campylobacter, Cryptosporidium or Giardia) and the presence of diarrhoeagenic E. coli-associated genes (stx and eae) were detected in 89.9% (179/199) samples, including 80.5% (33/41) of samples with putative non-recent faecal inputs. Quantitative PCR to detect microbial source tracking targets from human, ruminant and avian contamination were concordant with land-use type and E. coli phylotype abundance. This study demonstrated that a potential recreational health risk remains where pathogens occurred in water samples with low E. coli concentration, potential non-recent faecal sources, low impact sites and where human, ruminant and avian faecal sources were absent.