By use of the retrograde axonal transport of horseradish peroxidase (HRP), the projection from the superior colliculus (SC) to the brain stem reticular formation (RF) was investigated in the cat. A 0.2-0.5 microliter of a 50% suspension of Sigma VI HRP was injected stereotactically in various portions of the pontomedullary RF, and, as a control to the injection to the RF, in the inferior olive or in the spinal cord. Labeled cells were found within and deep to the intermediate gray layer of the SC in the cats which survived for two or three days after HRP injection. The number of the labeled cells varied, according to the difference in the site of injection and the amount of injected HRP. About 400 labeled cells in twenty 50-micron sections, taken every fifth of the SC, occurred throughout its rostrocaudal extent, particularly in the case where the medial portion of the border zone of the nucleus reticularis pontis oralis and caudalis (R.p.o.-R.p.c. zone) or the border zone of the nucleus reticularis pontis caudalis and the gigantocellularis (R.p.c.-R.gc. zone) was heavily stained after three days of survival period. From 10 to 15% of these labeled cells were large in size (more than 40 micron in diameter), 20-30% were medium sized and the rest (60-70%) were small (10-25 micron). On the other hand, when HRP was placed in the inferior olive only eight cells were labeled in the SC, seven of which were small and medium-sized. When HRP was injected in the gray matter of C1-C3 level of the spinal cord, a total of 70 tectal cells (14, 42, and 14 were large, medium, and small cells, respectively) were observed to be labeled. The findings of the tectoreticular neurons are discussed and compared with those of the tectoolivary and the tectospinal neurons. Thus the three kinds of tectal neurons are located within and deep to the intermediate gray layer. The number of the labeled cells and the percentages of the collicular neurons of different sizes are obviously different among the three different projections. Topographic correlations between the SC and the RF could not be discerned in the present materials. These results were discussed in relation to possible influences of the tectoreticular neurons upon the extraocular and the spinal motoneurons.