The developing uncrossed retinocollicular projection in the Syrian hamster undergoes a characteristic set of changes during the first 2 postnatal weeks. The retinal fibres, which initially project across the whole superior colliculus, withdraw from the caudal part and their terminals become clustered into deep, discrete clumps rostrally. Coincident with these afferent changes, there is substantial retinal ganglion cell death. To examine whether neuronal activity plays a role in these changes, we made daily injections of the sodium channel blocker tetrodotoxin (TTX) into one or both eyes from postnatal day 2 or 4 up to day 12. Following TTX treatment, the uncrossed terminals retracted on schedule from the caudal and superficial parts of the superior colliculus and came to lie, as normal, in the deep layers rostrally. Within the rostral superior colliculus, however, the uncrossed terminals from TTX-injected eyes lost their characteristic patchy distribution and were arranged diffusely. When only one eye received TTX injections, this inhibiting effect on terminal segregation was seen only in the projections from the TTX-treated eye. The effect of TTX treatment on terminal segregation was much less severe than that of unilateral enucleation, after which uncrossed terminals persis throughout the entire superior colliculus. TTX injections appeared to have little effect on overall ganglion cell death since the total number of ganglion cells in the crossed projection from TTX-treated eyes was similar to that in normal eyes. However, the relative distribution of uncrossed cells in temporal and nasal retina was altered. In eyes that received TTX injections, the proportion of uncrossed cells in nasal retina was about 1.6 times that in normal animals and was close to the proportion seen in unilaterally enucleated animals. This increase in the treated eye occurred whether one or both eyes were injected with TTX. We conclude that neuronal activity plays a role in the segregation of uncrossed terminals into discrete clumps in rostral colliculus and in the preferential elimination of uncrossed cells from the nasal retina. The inactive uncrossed projections from TTX-treated eyes showed the greatest degree of disruption. The extent of the disruption was similar whether the crossed input from the other eye was active or inactive. This suggests that the activity-drive interactions between ganglion cells within one eye are more significant than those between the two eyes in shaping the final form of the uncrossed retinocollicular projection.