Hepatitis C virus (HCV) infection is a major cause of chronic liver disease and has become a global health threat. No HCV vaccine is currently available and treatment with antiviral therapy is associated with adverse side effects. Moreover, there is no preventive therapy for recurrent hepatitis C post liver transplantation. The NS3 serine protease is necessary for HCV replication and represents a prime target for developing anti HCV therapies. Recently we described a therapeutic approach for eradication of HCV infected cells that is based on protein delivery of two NS3 protease-activatable recombinant toxins we named "zymoxins". These toxins were inactivated by fusion to rationally designed inhibitory peptides via NS3-cleavable linkers. Once delivered to cells where NS3 protease is present, the inhibitory peptide is removed resulting in re-activation of cytotoxic activity. The zymoxins we described suffered from two limitations: they required high levels of protease for activation and had basal activities in the un-activated form that resulted in a narrow potential therapeutic window. Here, we present a solution that overcame the major limitations of the "first generation zymoxins" by converting MazF ribonuclease, the toxic component of the E. coli chromosomal MazEF toxin-antitoxin system, into an NS3-activated zymoxin that is introduced to cells by means of gene delivery. We constructed an expression cassette that encodes for a single polypeptide that incorporates both the toxin and a fragment of its potent natural antidote, MazE, linked via an NS3-cleavable linker. While covalently paired to its inhibitor, the ribonuclease is well tolerated when expressed in naïve, healthy cells. In contrast, activating proteolysis that is induced by even low levels of NS3, results in an eradication of NS3 expressing model cells and HCV infected cells. Zymoxins may thus become a valuable tool in eradicating cells infected by intracellular pathogens that express intracellular proteases.