Intra-cerebral (i.c.) microinfusion of selective receptor agonists and antagonists into behaving animals can provide both neuroanatomical and neurochemical insights into the neural mechanisms of anxiety. However, there have been no systematic reviews of the results of this experimental approach that include both a range of unconditioned anxiety reactions and a sufficiently broad theoretical context. Here we focus on amino acid, monoamine, cholinergic and peptidergic receptor ligands microinfused into neural structures previously implicated in anxiety, and subsequent behavioral effects in animal models of unconditioned anxiety or fear. GABAA receptor agonists and glutamate receptor antagonists produced the most robust anxiolytic-like behavioral effects, in the majority of neural substrates and animal models. In contrast, ligands of the other receptor systems had more selective, site-specific anti-anxiety effects. For example, 5-HT1A receptor agonists produced anxiolytic-like effects in the raphe nuclei, but inconsistent effects in the amygdala, septum, and hippocampus. Conversely, 5-HT3 receptor antagonists produced anxiolytic-like effects in the amygdala but not in the raphe nuclei. Nicotinic receptor agonists produced anxiolytic-like effects in the raphe and anxiogenic effects in the septum and hippocampus. Unexpectedly, physostigmine, a general cholinergic agonist, produced anxiolytic-like effects in the hippocampus. Neuropeptide receptors, although they are popular targets for the development of selective anxiolytic agents, had the least reliable effects across different animal models and brain structures, perhaps due in part to the fact that selective receptor ligands are relatively scarce. While some inconsistencies in the microinfusion data can easily be attributed to pharmacological variables such as dose or ligand selectivity, in other instances pharmacological explanations are more difficult to invoke: e.g., even the same dose of a known anxiolytic compound (midazolam) with a known mechanism of action (the benzodiazepine-GABAA receptor complex), can selectively affect different fear reactions depending upon the different subregions of the nucleus into which it is infused (CeA versus BLA). These particular functional dissociations are important and may depend on the ability of a GABAA receptor agonist to interact with distinct isoforms and combinations of GABAA receptor subunits (e.g., alpha1-6, beta1-3, Upsilon1-2, delta), many of which are unevenly distributed throughout the brain. Although this molecular hypothesis awaits thorough evaluation, the microinfusion data overall give some support for a model of "anxiety" that is functionally segregated along different levels of a neural hierarchy, analogous in some ways to the organization of sensorimotor systems.