Many studies have been made on the distributions of CSF contacting neurons (CSF-CNs) in the parenchyma of the brain with horseradish peroxidase (HRP) or autoradiographics. A significant amount of data has shown that both HRP and autoradiographical substances could pass freely through the spaces of ependyma into the parenchyma of the brain. It is therefore possible that the results were not exact. We found that CB-HRP was a dependable tracer to CSF-CNs and studied the distributions and the signaling directions of cerebrospinal fluid contacting neurons (CSF-CNs) in the parenchyma of the brain with the cholera toxin subunit B with horseradish peroxidase (CB-HRP) tracing combined with transmission electron microscopy. The results were as follows: (1) CSF contacting tanycytes existed not only in the wall of the third ventricle (3V), but also in the walls of the lateral ventricle (LV), the fourth ventricle (4V) and the central canal (CC) of the spinal cord. (2) Some CSF contacting glia cells were observed in the lateral septal nucleus (LS). (3)The distal CSF-CNs in the parenchyma were found in LS, the anterodorsal thalamic nucleus (AD), the supramammillary nucleus (SuM), the dorsal raphe nucleus (DR), the floor of 4V and the lateral superior olive (LSO), but they were mainly found in DR and divided into groups A and B. (4) Axon terminals labeled by CB-HRP were found in the cavity of the brain ventricle. (5) The synaptic relationships between the neurons were labeled by CB-HRP in DR and no-labeled by CB-HRP in the parenchyma. Both synapses Gray I and II were found. It was significant that the presynaptic elements were formed by the neurons no-labeled CB-HRP and the postsynaptic elements labeled CB-HRP. Our results suggested firstly that the signaling directions of CSF-CNs in DR were only from the parenchyma to CSF.