To study the developmental regulation of a neuropeptide phenotype, we have analyzed the biochemical and morphological differentiation of two identifiable neurons in embryos of the moth, Manduca sexta. The central cell, CF, and the peripheral cell, L1, are both neuroendocrine neurons that express neuropeptides related to the molluscan tetrapeptide FMRFamide. Both neurons project axons to the transverse nerve in each thoracic segment. Within the CF and L1 cells, neuropeptide-like immunoreactivity was localized to secretory granules that had cell-specific morphologies and sizes. The onset of neuropeptide expression in the two cell types displayed a similar pattern: immunoreactivity was first detected in distal processes and soon after within cell bodies. However, the onsets occurred at different times: for the CF cell, neuropeptides were first seen at 60%-63% of embryonic development, after the neuron had extended a long axon into the periphery, while L1 neuropeptide expression began at approximately 42%, as it first extended its growth cone. These times were related in that they corresponded to the arrival times of the respective growth cones at a similar position in the developing peripheral nerve. Within this region of the nerve, the growth cones of both cell types-exhibited a transient and cell-specific interaction with an identified mesodermal cell, called the Syncytium. Like the L1 and B neurons (Carr and Taghert, 1988b), the CF growth cones typically grew past this cell, yet remained attached to it by lamellipodial and filopodial processes of the axon. Ultrastructurally, the interaction involved filopodial adhesion to and insertion within the Syncytial cell. Two other nonneuroendocrine cell types grew axons past this same region, but showed no such tendencies. To test the hypothesis that the morphological and biochemical differentiation of these cells was somehow linked, central ganglia were isolated (as individuals or connected as ganglionic chains) in tissue culture, prior to the time when CF growth cones entered the periphery and prior to the development of CF neuropeptide expression. In the majority of cases, CF neurons nevertheless displayed their neuropeptide phenotype at a normal and cell-specific stage. We conclude that the initiation of neuropeptide expression is highly correlated with schedules of morphological differentiation in these neurons, but that, in the case of the CF neuron, it is not regulated by interactions of the growth cone with peripheral structures.