Ascending infection of the upper female genital tract with Gram-negative bacteria causes pelvic inflammatory disease (PID) or endometritis in women, with the influx of neutrophils and macrophages leading to accumulation of pus in the uterine lumen. Gram-negative bacteria are well adapted to colonise the human female genital tract and disease models have been established in mice. Pelvic inflammatory disease is also common in cattle because the uterine lumen of almost all animals is contaminated by a wide range of bacteria after parturition. The most important bacteria in the uterus are E. coli and it was assumed that random strains from the environment infect the uterus. However, when we tested this assumption, we were surprised to find that PID was associated with specific strains of E. coli. Using several molecular genotyping techniques we confirmed that these strains of E. coli were distinct from the well-known strains of E. coli that cause other diseases in animals or humans. In the present study we showed that the distinct disease-associated bacteria also appeared to be adapted to the endometrium. They were adherent and invasive for endometrial cells and stimulated the secretion of inflammatory mediators. The inflammation was associated with bacterial lipopolysaccharide (LPS) acting via the specific LPS receptor Toll-like receptor 4 (TLR4) but the E. coli did not express the expected virulence genes associated with other pathogenic E. coli. The uterine bacteria or their LPS also established pelvic inflammatory disease in mice, providing a useful model for research of mucosal infection and immunity. So, it appears that we have discovered specific strains of E. coli that cause disease in the uterus. Researchers developing vaccines or treatments for pelvic inflammatory disease should now focus on these new specific strains of E. coli.