The contribution of activity to the long-term stability of synaptic connections is a subject of ongoing debate. In the present study we examined the effects of recovery from chronic disuse on the pattern of reinnervation of partially denervated adult rat skeletal muscles, using tension measurements, intracellular recordings, and observations of vital staining with activity-dependent styryl dyes. Fourth deep lumbrical muscles were partially denervated by crushing the lateral plantar nerve (LPN) bilaterally. Denervated muscle fibers became innervated by sprouts from the sural nerve (SN); 17-23 d after LPN crush, conduction in the right sciatic nerve was blocked by connecting an osmotic minipump containing tetrodotoxin to a cuff placed around the nerve. Distal muscles remained continuously paralysed for 10-19 d. After 2 weeks of nerve block the tension produced by stimulating the regenerated LPN axons had reached about 70% of the total. Regenerating axons in contralateral muscles reinnervated only about 55% of the muscle fibers. The level of dual innervation by both regenerating and intact axons reached about 50% of the total muscle fibers after 2 weeks of paralysis, but only about 20% in contralateral controls. We then measured the rate and amount of decline in motor unit tension and polyneuronal innervation in reinnervated muscles during an eight week period of recovery from nerve block. Some LPN and SN connections regressed within 2-4 weeks, but about 35% of the muscle fibers still retained convergent inputs from LPN and SN motor axons 8 weeks after activity had resumed. This was about twice the level observed in contralateral reinnervated muscles. Vital staining with the styryl dyes FM1-43 and RH414 confirmed that many of the reinnervated motor end-plates were convergently supplied by both SN and LPN axons. Intracellular recordings showed that most of the dually innervated fibers in paralyzed muscles were supplied by suprathreshold inputs from both LPN and SN axons. The increased excitability of these muscle fibers was partly explained by their two-fold increased input resistance. Input resistance recovered to control levels within 4 weeks of resumption of activity, but dually innervated fibers in previously blocked muscles still gave mostly suprathreshold responses to stimulation of both the LPN and the SN. We conclude that chronic nerve conduction block generates or sustains a local environment which allows some convergent synaptic inputs on reinnervated muscle fibers to become consolidated and strengthened, independent of subsequent neuromuscular activity.