Despite longstanding promise and many known examples, antimicrobial peptides (AMPs) have failed, with few exceptions, to significantly impact human medicine. We have demonstrated that host cell interactions can contribute significantly to the loss of AMP activity, in vivo, where host cells are highly concentrated, through direct binding and through degradation. Indeed, we show that most of a representative set of natural and synthetic AMPs tested are severely inhibited by human erythrocytes and would be effectively inactive at physiological host cell concentrations. To circumvent these impediments, we developed a screen for sterilizing antibacterial activity against multiple organisms in the presence of concentrated human erythrocytes. We used the screen to select the most active peptides from a second generation peptide library based on a known AMP. The resulting peptides are potent against many bacteria, show no host-cell dependent inhibitions, and have essentially zero toxicity. Furthermore, we have shown that the lead peptides do not induce any resistance in bacteria under conditions where they rapidly become resistant to conventional antibiotics and other AMPs. The new AMPs we have discovered are thus are uniquely well suited to retain systemic antibiotic activity in vivo.