During the last few years, the use of type III secretion system-based bacterial vectors for immunotherapy purposes has been assessed in various applications. We showed that a type III secretion-based Pseudomonas aeruginosa vector delivering the ovalbumin (OVA) antigen induced an efficient specific CD8 ؉ T-lymphocyte immune response against OVA-expressing cells. Because of the intrinsic toxicity of the vector, further virulence attenuation was needed. Therefore, we explored the effects of the deletion of quorum-sensing genes and the aroA gene toward toxicity and efficiency of the vector strain. The aroA mutation in our strain (making the strain auxotrophic for aromatic amino acids) conferred a strikingly reduced toxicity, with the bacterial lethal dose being more than 100 times higher than that of the parental strain. The quorum-sensing gene mutation alone was associated with a slightly reduced toxicity. In a prophylactic OVA-expressing melanoma mouse model, an OVA-delivering aroA-deficient mutant was the most efficient at a low dose (10 5 ), but dose enhancement was not associated with a greater immune response. The quorum-sensing-deficient strain was the most efficient at a mild dose (10 6 ), but this dose was close to the toxic dose. Combination of both mutations conferred the highest efficiency at an elevated dose (10 7 ), in agreement with the known negative effects of quorum-sensing molecules upon T-cell activation. In conclusion, we have obtained a promising immunotherapy vector regarding toxicity and efficiency for further developments in both antitumor and anti-infectious strategies.The use of live bacteria and bacterial virulence factors as therapeutic tools in human medicine has been considered for more than a century. The observation that the onset of a bacterial infection could modify the course of a malignant disease (6) was a hallmark in this history, but in the end very few procedures (such as the intravesical administration of an attenuated Mycobacterium bovis strain for the cure of noninvasive urothelial carcinoma) have been routinely used. In the last 10 years, better characterization of bacterial mechanisms (mainly toxins and secretion systems) and extensive progress in genomic studies have allowed engineering of bacteria (mainly Escherichia coli and Salmonella spp.). These domesticated agents can be delivered to mammals for different purposes. Notably, the design of antigen-delivering bacteria that trigger antigen-specific cytotoxic CD8 ϩ T-lymphocyte responses is an emerging field of investigation in vaccine development (5). Antigen delivery can be performed by using intrinsic properties of bacterial toxins (as with a Listeria monocytogenes-derived vector [21]) or secretion pathways normally used by bacteria to release toxins, such as the type III secretion system (TTSS). This system has been considered promising because it allows gram-negative rods to inject toxins into eukaryotic cell cytoplasm; therefore, epitopes delivered by this system are likely to be presented by antigen-presenting c...