Remodeling of the synaptic extracellular matrix (ECM) and its dynamic relationship with nerve terminal plasticity have been demonstrated in normal frog neuromuscular junctions (NMJs) in vivo (Chen et al., 1991). Our previous work has led to a hypothesis that extension of synaptic ECM precedes nerve terminal growth during synaptic remodeling. To test this hypothesis, the present study examined the changes of synaptic ECM in frog NMJs that were primarily undergoing nerve terminal growth and sprouting. Frog sartorius muscles were double stained with a fluorescent nerve terminal dye (4-Di-2-Asp) and rhodamine-tagged peanut agglutinin (PNA), which recognizes synaptic ECM. The double-labeled NMJs were visualized in vivo with video-enhanced fluorescence microscopy. Nerve sprouting was then induced in the muscle by grafting segments of the contralateral sciatic nerve. The identified NMJs were restrained and reexamined 2–3 months later. Extensive sprouting was observed in 46% of 167 identified NMJs. At junctional regions that showed extension or formation of new branches, synaptic ECM was commonly seen to have the same shape and distribution as the nerve terminal. However, extension of synaptic ECM beyond the corresponding nerve terminals, often by tens of microns, was observed in 29% of these newly formed junctional regions. This lack of correlation might be transient, as growth of nerve terminals following extended, PNA-stained ECM was seen. Examination with histological staining not only confirmed a lack of nerve terminal at the extended synaptic ECM region but also indicated an absence of AChE and postsynaptic junctional folds. The absence of these postsynaptic specializations at the extended, PNA- stained ECM region makes it unlikely that this region was previously occupied by nerve terminals that had retracted. Thus, the present study provides further findings consistent with the hypothesis that synaptic ECM precedes nerve terminal outgrowth and that the extension of synaptic ECM may play a role in synaptic remodeling.