1. We studied the morphology of individual, physiologically identified retinogeniculate axons in normal adult cats. The axons were recorded in the lateral geniculate nucleus or in the subjacent optic tract, characterized as X or Y by physiological criteria, penetrated, and injected with horseradish peroxidase. With subsequent application of appropriate histochemistry, the enzyme provides a complete label of the terminal arbors and parent trunks for morphological analysis. We have recovered for such analysis 26 X- and 25 Y-axons; of these, 14 X- and 12 Y-axons were studied in detail. 2. Within the optic tract, the parent trunk of every X-axon is located closer to the lateral geniculate nucleus and thus further from the pial surface than that of every Y-axon. This probably reflects the earlier development of X- than of Y-axons. Furthermore, the parent axon trunks of the X-axons are noticeably thinner than are those of the Y-axons. Every retinogeniculate X- and Y-axon in our sample branches within the optic tract. One of these branches heads dorsally to innervate the lateral geniculate nucleus and one heads medially and rostrally toward the midbrain, although none of these labeled axons were traced to a terminal arbor beyond the lateral geniculate nucleus. For Y-axons, all branches are of comparable diameter, but for X-axons, the branch heading toward the lateral geniculate nucleus is always noticeably thicker than is the branch directed toward the midbrain. 3. Every retinogeniculate X- and Y-axon produces the greatest portion of its terminal arbor in lamina A (if from the contralateral retina) or A1 (if from the ipsilateral retina). These arbors typically extend across most of the lamina along a projection line. Not a single terminal bouton from any axon was found in the inappropriate lamina A or A1 (i.e., in lamina A for ipsilaterally projecting axons or in lamina A1 for contralaterally projecting ones). Occasionally, an X-axon also innervates the medial interlaminar nucleus, and even more rarely does an X-axon innervate the C-laminae. In contrast, nearly all Y-axons from the contralateral retina branch to innervate part of the C-laminae (probably lamina C), and most from either retina also innervate the medial interlaminar nucleus. Although these details imply considerable variation in the overall pattern of retinogeniculate innervation for both X- and Y-axons, we found no physiological properties to correlate with this variation. 4. Within lamina A or A1, certain differences were noted between X- and Y-axon arbors. Compared with Y-arbors on average, the X-arbors are half as wide in the mediolateral direction (150 μm vs. 300 μm) and one-quarter the volume, and they produce only about half as many terminal boutons (500-600 vs. 1,000). Although a wide range of bouton sizes (2-6 μm diam) is evident for each arbor, Y-axons contain a wider range of sizes with many more of the larger ones than do X-axons. Finally, the boutons in X-arbors are found on short stalks in prominent clumps, whereas those in Y-arbors are more diffusely distributed en passant along the preterminal axon branches. 5. In addition to forming more boutons than do X-axons in laminae A and A1, Y-axons also form many boutons elsewhere in the lateral geniculate nucleus, whereas such boutons beyond the A-laminae are rare for X-axons. Each Y-axon thus seems able to innervate more geniculate neurons, which may represent the morphological basis for previously published evidence that the X-to-Y cell ratio is considerably lower in the lateral geniculate nucleus than in the retina. A similar difference between X- and Y-axons has already been described for geniculocortical innervation. This pattern of greater synaptic numbers for individual Y- than for X-axons may serve to explain in part how the Y pathway, which is so small at its retinal origin, can become so important at the level of visual cortex.