This dissertation thesis unravels various strategies of the Drosophila melanogaster olfactory circuit to decode the olfactory world into an accurate perception of odors underlying innate odor-guided behavior. The olfactory system is an extremely flat but elegantly organized circuitry, which is optimized to extract relevant variables at any of its processing levels. The peripheral machinery composed of various OSN types mainly serves as a pure detection area encoding odors in a combinatorial manner. Already at the first synapse between OSNs and PNs specific features of an odor stimulus as valence and intensity are extracted by the PNs that relay this information to higher processing centers. Hereby excitatory PNs, which represent the majority of the AL output, possess the capacity to extract odor identity as well as positive and negative valence. Inhibitory PNs constitute a parallel processing stream to ePNs towards the higher brain and perform a decoding of features from the AL into a particular percept underlying a traceable behavior. Moreover, I found evidence for a spatial partitioning of the LH according to distinct odor features. Since both PN types are reunited within the LH, valence and intensity-specific activity might be computed to configure a valence-specific behavioral output. Besides general rules of olfactory coding, this thesis elucidates two exceptional cases with evolutionarily highly conserved underlying principles. First, the olfactory mimicry of yeasty fermentation products performed by the plume of the deceptive Arum palaestiunum flower and second, a labeled line circuit of fundamental significance for the animal’s survival could be discovered: the geosmin pathway is exclusively tuned to the repellent product of toxic microorganisms. In conclusion, findings within my thesis reveal general rules for decoding odor features as innate valence and odor intensity as well as specialized hard-wired mechanisms initiating innate behavior.