During development, chick lumbosacral motoneurons have been reported to form precise topographic projections within the limb from the time of initial outgrowth. This observation implies, first, that motoneurons select the appropriate muscle nerve pathway and, second, that they restrict their ramification within the primary uncleaved muscle masses to appropriate regions. Several reports based on electrophysiology and orthograde horseradish peroxidase (HRP) labeling have shown muscle nerve pathway selection to be fairly precise. However, studies based on retrograde labeling with HRP have produced conflicting reports on the extent to which vertebrate motoneurons make projection errors. Since it is difficult to distinguish between true projection errors and HRP leakage when using retrograde labeling, we decided to assess the distribution of labeled growth cones in 25-micron serial plastic sections, following orthograde labeling of identifiable subpopulations of motoneurons during the period of initial axon outgrowth. Examination of a large number of muscle nerves revealed no segmentally inappropriate axons, confirming earlier reports that muscle nerve pathway selection is very accurate. In addition, we observed that growth cones take widely divergent trajectories into the same muscle nerve, suggesting that growth cones are responding independently to some specific environmental cue rather than being passively channeled at this point. The distribution of labeled growth cones within the muscle masses provided direct evidence that motoneurons did not at any time project to obviously inappropriate muscle regions.(ABSTRACT TRUNCATED AT 250 WORDS)