Few studies have focused on the transformation of growth cones to mature synaptic arbors. To study these events in developing axons in vivo, we have labeled growing cerebellar axons with horseradish peroxidase (HRP) in postnatal stages [Mason and Gregory, 1984]. This report will provide the first data on embryonic cerebellar axons, and will ask whether growth cones differ in tracts and in target tissue and what features characterize axons that enter the cerebellum in fetal periods. During the earliest embryonic (E) periods examined (E16-19), axons in tracts have enlarged growth cones with lamellopodia and short filopodia that contain small and large vesicles. In contrast, axons within the cerebellar anlage from E16-postnatal day (P) 5 have fine calibers with a minimum of branching, and have small tapered growing tips. If synaptic contacts are made by such growing tips, there is little concomitant change in their shape. As target cells from layers and as their dendrites extend (P5-P7), growing tips and synaptic boutons differentiate according to the type of synaptic arrangement in which they engage. Enlarged, irregular expansions of growing tips correspond to synaptic contacts with multiple dendritic partners and are filled with large and small clear vesicles. Filopodia arising from such swellings, like the small undifferentiated growing tips of the type seen on embryonic axons, contain a mixture of vesicle types but make simple synapses with single profiles. Many axons make both kinds of synaptic structures, especially during the period when maturing axons give rise to long filopodia. Thus, growing tips have immature forms long after synaptogenesis begins, and use filopodial structures to elaborate synaptic arrangements. This analysis should elucidate the changes in growth cone form and cytology that reflect cell-cell interactions during synaptogenesis.