Collateralization of individual cortical axons is well documented for rodents but less so for monkeys, where double retrograde tracer experiments have tended to find only small numbers of neurons projecting to two different injection sites. Evidence from both double label and single axon studies, however, suggests that in specific projection systems the number of neurons with collateralized axons can be 10% or greater. These include feedback projections from temporal areas (but less so those from V4 and MT/V5). Single-axon analyses show that many parietal neurons branch to multiple targets. Except for giant Meynert cells in area V1, feedforward projections from early visual areas have only a small number of neurons with branching axons. Why only some neurons collateralize, what determines branch points and projection foci, and how this impacts network organization are largely unknown. Deciphering the branching code might offer new perspectives on space-time organization at the network level.