In this study, we investigated the influence of semantic and perceptual learning on the perception of odor blending mixtures, i.e., mixtures eliciting a different odor quality as compared with their components. Following a between-groups design, we set out to examine whether being pre-exposed to target mixture odor labels (semantic learning) and to the mixture's odor components (perceptual olfactory learning) would modify the perception of odor mixtures quality. In a first experimental condition, a group (control) of 28 subjects evaluated the odor quality of mixtures and then their components. In this condition, subjects were not exposed to the individual odor components (no perceptual learning) and were only weakly exposed to the target mixture odor labels (weak semantic learning) when asked to evaluate the quality of the mixtures. In a second experimental condition, a group (learning) of 29 other subjects performed the same tasks but evaluated first the components and then the mixtures. In this condition, subjects have experienced both perceptual and strong semantic learning when asked to evaluate the quality of the mixtures. Within each condition, odor quality was evaluated following two verbal tasks: a free description task and a choice between several odor attributes. The comparison of the data obtained in both experimental conditions revealed that semantic and perceptual learning could influence odor mixture quality and suggested a differential influence of both learning types according to mixture complexity.