Using a vital nerve terminal dye (4-Di-2-ASP) and fluorescently tagged alpha-bungarotoxin to stain postsynaptic acetylcholine (ACh) receptors, we viewed the same muscle fibers at multiple times in the sternomastoid muscle of living mice during the process of reinnervation following nerve crush. Soon after axons reenter the muscle, they precisely reoccupy the original endplate sites. However, in contrast to normal adult muscle, during the first several weeks of reinnervation, anatomical and physiological measures show that many of the endplate sites are innervated by more than one axon. Typically, one axon reinnervates the original endplate site by growing up the old Schwann cell tube while another originates as a sprout from a nearby endplate. Within 2 weeks after reinnervation nerve terminal staining shows that most of the sprouts have regressed and physiological evidence of multiple innervation has returned to the normal low level. By repeatedly observing the same endplates during the period of synapse elimination, we could directly view this phenomenon. At some endplates, nerve terminal boutons in one region of the endplate were eliminated at the same time a sprout entering that area regressed. These unoccupied sites seemed permanently eliminated as they are not subsequently occupied by sprouts from the axon remaining at the endplate. We were surprised to find that there is a corresponding permanent loss of ACh receptors within the muscle fiber membrane precisely underneath the eliminated nerve terminals. The decrease in receptors at sites of synapse elimination is due to both a selective loss of ACh receptors already incorporated into these sites and to a lack of insertion of new receptors at the same regions. These sites of pre- and postsynaptic loss, however, maintain cholinesterase staining in the basal lamina for long periods. Control experiments showed that endplates that were permanently denervated, incompletely reoccupied by reinnervating axons, or stained and viewed multiple times in normal muscle do not lose postsynaptic receptor regions. Interestingly, receptors appear to be eliminated before there is any obvious change in the staining of the overlying nerve terminal. Because of the lag between receptor and nerve terminal loss, we could predict which synaptic boutons would be eliminated by looking for lightly stained receptor regions. One interpretation of these data is that the removal or redistribution of relevant postsynaptic molecules by one innervating axon may instigate the elimination of competing terminals.