AbstractEscherichia coli is the leading cause of urinary tract infection, one of the most common bacterial infections in humans. Despite this, a genomic perspective is lacking regarding the phylogenetic distribution of isolates associated with different clinical syndromes. Here, we present a large-scale phylogenomic analysis of a spatiotemporally and clinically diverse set of 907 E. coli isolates, including 722 uropathogenic E. coli (UPEC) isolates. A genome-wide association approach identifies the (P-fimbriae-encoding) papGII locus as the key feature distinguishing invasive UPEC, defined as isolates associated with severe UTI, i.e., kidney infection (pyelonephritis) or urinary-source bacteremia, from non-invasive UPEC, defined as isolates associated with asymptomatic bacteriuria or bladder infection (cystitis). Within the E. coli population, distinct invasive UPEC lineages emerged through repeated horizontal acquisition of diverse papGII-containing pathogenicity islands. Our findings elucidate the molecular determinants of severe UTI and have implications for the early detection of this pathogen.