Clinically relevant multidrug-resistant bacteria often arise due to overproduction of membrane-embedded efflux proteins that are capable of pumping antibiotics out of the bacterial cell before the drugs can exert their intended toxic effect. The Escherichia coli membrane protein AcrB is the archetypal protein utilized for bacterial efflux study because it can extrude a diverse range of antibiotic substrates and has close homologues in many Gram-negative pathogens. Three AcrB subunits, each of which contains 12 transmembrane (TM) helices, are known to trimerize to form the minimal functional unit, stabilized noncovalently by helix–helix interactions between TM1 and TM8. To inhibit the efflux activity of AcrB, we have rationally designed synthetic peptides aimed at destabilizing the AcrB trimerization interface by outcompeting the subunit interaction sites within the membrane. Here we report that peptides mimicking TM1 or TM8, with flanking N-terminal peptoid tags, and C-terminal lysine tags that aid in directing the peptides to their membrane-embedded target, decrease the AcrB-mediated efflux of the fluorescent substrate Nile red and potentiate the effect of the antimicrobials chloramphenicol and ethidium bromide. To further characterize the motif encompassing the interaction between TM1 and TM8, we used Förster resonance energy transfer to demonstrate dimerization. Using the TM1 and TM8 peptides, in conjunction with several selected mutant peptides, we highlight residues that may increase the potency and specificity of the peptide drug candidates. In targeting membrane-embedded protein–protein interactions, this work represents a novel approach to AcrB inhibition and, more broadly, a potential route to a new category of efflux pump inhibitors.