We sought to establish what proportion of the cholinergic innervation of the cerebral cortex (CX) is associated with intraparenchymal blood vessels by using immunocytochemical and neurochemical techniques, and whether [ 3H]acetylcholine ([ 3H]ACh) is synthesized and released by elements associated with cortical microvessels (MV). MVs and, for comparison, tissue homogenates were prepared using sucrose gradient/differential ultracentrifugation methods. Efficacy of the separation technique was indicated by the activity of γ-glutamyltranspeptidase (up to 29.2-fold enrichment), an endothelial cell marker enzyme, in the MV fraction and microscopy. The size of isolated microvessels ranged from 5 to 40 μm (o.d.) with 67.7% of the vessels less than 10 μm and 32.2% between 11 and 40 μm (690 vessels measured from 4 animals). By electron microscopy immunoreactive choline acetyltransferase (ChAT), the biosynthetic enzyme for ACh, was localized to: (a) axons and axon terminals opposed to the basal laminae of capillaries and small arterioles, and (b) capillary endothelial cells. ChAT-labeled elements associated with MVs were most prominent in layers I, III and V of the CX consistent with the local pattern of cholinergic innervation. The absolute amount of ACh synthesized (pmol Ach/100 mg wet wt.) by elements associated with cortical MVs was relatively small (2.3% total cortical homogenate activity). Inhibition of MV ChAt activity to 5% of control by the specific ChAT inhibitor, 4-naphthylvinylpyridine, and HPLC analysis of the product, indicated that authentic ACh was measured. Other tissues similarly synthesized small amounts of ACh relative to the CX, caudate nucleus (CN, 2.4%), cerebellum (CRB, 1.4%) and liver (LIV, 3.9%). Consistent with the known extent of the cholinergic innervation of the tissues examined, the rank order of ChAT associated for both MVs and homogenate were: CN > CX > > CRB > LIV. However, based on the specific activities of ChAT, cortical MVs have the remarkable capacity to synthesize ACh at rates 95% greater than cortical (S 1 fraction) homogenate (59.0 ± 3.5nmol/mg protein/40min; n = 7), which is enriched in nerve terminals. Except for LV (+11%), other tissues also had remarkably high ChAT activity in MV (% above corresponding homogenate; P < 0.05, n = 5): CN (+269) and CRB (+313). Release of [ 3H]ACh from MVs and, for comparison, nerve terminals were graded to K + depolarization stimulus (5–55 mM), maximal with 55 mM K + and CA 2+ dependent. The K +-evoked release of neurotransmitter amino acids aspartate and γ-aminobutyric acid (GABA), unlike [ 3H]ACh, was only observed from nerve terminals. This differential pattern of neurotransmitter release suggest a selective innervation of cholinergic neurons with the cortical microvasculature and that contamination of the MV fraction by non-vascularly related neurons in unlikely. We conclude that the synthesis and release of ACh, at the level of cortical MVs, is consistent with evidence for a potent mechanism for the neural control of the cerebral circulation by ACh.