During normal postnatal maturation, mammalian muscles undergo an orderly process of synapse elimination, whereby each muscle fibre loses all but one of the multiple inputs with which it is endowed at birth. Experimental perturbations that increase or decrease the overall activity of nerve and/or muscle cause a corresponding increase or decrease in the overall rate of neuromuscular synapse elimination. On other grounds it has been suggested that competition among motor neurons is important in determining which synapses survive and which are eliminated. Would a difference in activity among the terminals at the same endplate affect the outcome of the competition and not just its rate? We investigated this issue by blocking activity for four days in a small fraction of the motor neurons innervating the neonatal rabbit soleus muscle. Twitch tensions of motor units were subsequently measured for both the active and inactive populations of neurons to assess whether the inactive neurons had lost fewer or more synapses than is normal. We found that inactive motor neurons have a significant advantage compared to active counterparts in control experiments, a finding opposite to that expected if the neuromuscular junction operated by classical 'Hebbian' rules of competition.