Following the injection of horeseradish peroxidase (HRP) into the brachial spinal cord of the chick on embryonic day (E)4.5, retrogradely labeled neurons can be found in the brainstem (Okado and Oppenheim: Journal of Comparative Neurology 232: 143-161, 1985). By contrast, following high cervical spinal transection, functional (behavioral) deficits are not observed until E10 (Oppenheim: Journal of Comparative Neurology 160: 37-50, 1975). To determine whether this temporal difference between projections and function reflects a delay in synaptogenesis, we looked for the presence of anterogradely HRP-labeled pre-synaptic terminals in brachial cord following injection of HRP into the boundary between brainstem and spinal cord at ages between E3.5 and E7. HRP-labeled fibers were observed in the branchial cord by E4.5 and were diffusely distributed in the ventral and lateral marginal zones (presumptive ventral and lateral funiculi, respectively). Although some axo-dendritic and axo-somatic synapses were observed in the brachial cord prior to E6, the presynaptic profiles were always unlabeled by HRP and thus must originate from propriospinal sources. The first HRP-labeled supraspinal synapses were found in the ventral and lateral funiculi on E6. They contained several clear spherical synaptic vesicles and were axo-dendritic in nature. The cells of origin of the postsynaptic dendrites were determined by injecting HRP into the wing-bud to label the brachial motoneurons retrogradely and the presynaptic component was identified as supraspinal by HRP injections into the brainstem/spinal cord boundary to orthogradely label the descending fibers. Several double-labeled axo-dendritic synapses were found in the ventral and lateral funiculi of E6 brachial cord. Therefore, at least some descending supraspinal fibers make synapses directly onto motoneuron dendrites. We conclude that 1) there is a delay of about 1.5 days between the arrival of supraspinal fibers and synapse formation in the brachial cord, 2) the earliest synapses are axo-dendritic in nature, 3) at least some supraspinal fibers make direct contact with motoneuron dendrites as early as E6, and 4) synaptogenesis from propriospinal sources precedes that from supraspinal descending axons. These observations provide evidence indicating that the temporal difference between the onset of projections of supraspinal descending fibers and the onset of their function may be partly owing to delayed synaptogenesis.