To identify mechanisms that regulate neuronal form in the mammalian CNS, we have examined dendritic development in the lateral geniculate nucleus (LGN) during the period of segregation of retinal ganglion cell axons. The tracer Dil was used to label retrogradely LGN neurons that send their axons to primary visual cortex at different ages between embryonic day 36 (E36) and E60 in the cat. LGN neurons grow extensively during this period, in concert with the progressive restriction of ganglion cell axons from the two eyes to their appropriate eye-specific layers. At E36 neurons have simple bipolar morphology; by E60 all have acquired complex multipolar dendritic trees. During this period, soma size increases by 190% and total dendritic length increases 240%. Dendritic complexity, as measured by dendritic branch points, also increases. As dendrites grow, the number of spines increases, but their density remains constant at 0.015/micron throughout this period. Since it is known that blockade of action potential activity significantly alters the branching pattern and extent of retinal ganglion cell axonal arbors within the LGN, we also investigated whether the dendritic development of the postsynaptic LGN neurons is similarly susceptible. Following 2 weeks of the intracranial minipump infusion of TTX between E42 and E56, the morphology of LGN neurons was examined. Surprisingly in view of the striking effect of the treatment on the morphology of retinal ganglion cell axons, dendritic growth and development were essentially normal. However, the density of dendritic spines increased almost threefold, suggesting that this specific feature of dendritic morphology is highly regulated by action potential activity. These observations indicate that normally during this period of development, the previously described changes that occur in the morphology of the presynaptic inputs to LGN neurons are accompanied by a progressive growth of post-synaptic dendrites. Because the intracranial TTX infusions have almost certainly blocked all sodium action potentials, our results suggest that the basic dendritic framework of LGN neurons can be achieved even in the absence of this form of neural activity. Moreover, since the same treatment causes a profound change in the morphology of the presynaptic axons, at least some aspects of axonal and dendritic form must be controlled independently during this prenatal period of development.