Previous work suggested that virulent bacteria, which can grow rapidly in serum, must possess a specific mechanism for removing iron from its transferrin complex. Two strains of Escherichia coli were examined with this in mind. Strain O141, which showed inoculum-dependent growth in serum and multiplied in the mouse peritoneum, secreted iron-binding catechols into both synthetic medium and serum. One of these compounds has an association constant for iron similar to that of transferrin. Both transferrin and ethylenediamine-di-o-hydroxyphenyl acetic acid (EDDA), which have very high affinities for ferric iron, induced catechol synthesis in growing cultures of strain O111. This organism was inhibited by normal horse serum. Further work showed that traces of specific antibody inhibited catechol synthesis by O111 exposed to EDDA; therefore, the existence of this inhibitory process means that the organism can no longer obtain Fe(3+), which all remains bound to transferrin in serum. In vivo, the inhibition of O111 is similar to that produced by serum in vitro. Neither phagocytosis nor killing by complement appeared to be of any significance during the first 4 h of the infections. Significantly, the purified catechol was capable of abolishing bacteriostasis in vivo. Since these results show that the production of iron-binding catechols is essential for rapid bacterial growth both in vitro and in vivo, these compounds should therefore be considered as true virulence factors. Conversely, any interference by the host with the production or activity of these compounds would constitute an important aspect of antibacterial defense.