A total of 208 single neurons were extracellularly recorded in the locus coeruleus (LC) of 11 cats. In later histofluorescence studies, greenish fluorescent LC neurons, from which we believed to have recorded well-isolated action potentials, were always found in or close to the center of red fluorescent halo due to an injected dye which marked the recording site. One hundred twelve of these 208 neurons were further subjected to electrical stimulation of the dorsal bundle of ascending axons originating from the norepinephrine (NE)-containing LC neurons and thus activated antidromically with a mean latency of 8.9 ms (the remaining neurons were lost before this examination). The mean conduction velocity was 1.2 m/s. Furthermore, it is suggested that 22% of thus antidromically identified NE neurons in the cat LC had an ascending axon of the conduction velocity faster than 2.4 m/s. This finding may be related with an electron-microscopic observation which indicated the presence of myelinated catecholamine (CA) axons, though not many in number, in the cat visual cortex. Responses by the NE-containing LC neurons to various natural visual stimuli, such as flashlight, moving and stationary light-slit, multiple spots, and gratings were examined. It turned out, however, that flash alone was effective to activate LC neurons. The latency of flash evoked activity was between 48 and 96 ms (N = 12; mean: 60 ms). Furthermore, the following areas in the central visual pathway were electrically stimulated to activate LC neurons orthodromically: the optic chiasm (OX), the dorsal lateral geniculate nucleus (LGN), the superior colliculus (SC), and the visual cortex (VC). The range and the mean of the latency for orthodromic responses were as follows: OX (N = 36, 8.4-42 ms; mean: 21 ms); LGN (N = 17, 6.0-17 ms; mean: 8.1 ms); SC (N = 12, 3.6-12 ms; mean: 5.6 ms); VC (N = 10, 7.8-40 ms; mean: 16.4 ms). The long latency of these orthodromic responses and its wide distribution suggest that afferents to the LC from the above-mentioned visual structures are most likely polysynaptic in nature. The extensive input convergence, including acoustic and nociceptive afferents, and the polysynaptic connection in each afferent pathway indicated a strong similarity between the afferent connectivity of NE-containing LC neurons revealed in the present study and that known for reticular formation neurons. Then, we would like to suggest that visual signals from the eyes impinge upon the NE-containing LC neurons via the reticular formation and that the afferents from the LGN, the SC, and the VC also join this common path through the reticular formation to reach the LC.