Several studies using transgenic mouse models of familial amyotrophic lateral sclerosis (ALS) have reported a life span increase in exercised animals, as long as animals are submitted to a moderate-intensity training protocol. However, the neuroprotective potential of exercise is still questionable. To gain further insight into the cellular basis of the exercise-induced effects in neuroprotection, we compared the efficiency of a swimming-based training, a high-frequency and -amplitude exercise that preferentially recruits the fast motor units, and of a moderate running-based training, that preferentially triggers the slow motor units, in an ALS mouse model. Surprisingly, we found that the swimming-induced benefits sustained the motor function and increased the ALS mouse life span by about 25 days. The magnitude of this beneficial effect is one of the highest among those induced by any therapeutic strategy in this disease. We have shown that, unlike running, swimming significantly delays spinal motoneuron death and, more specifically, the motoneurons of large soma area. Analysis of the muscular phenotype revealed a swimming-induced relative maintenance of the fast phenotype in fast-twitch muscles. Furthermore, the swimming programme preserved astrocyte and oligodendrocyte populations in ALS spinal cord. As a whole, these data are highly suggestive of a causal relationship not only linking motoneuron activation and protection, but also motoneuron protection and the maintenance of the motoneuron surrounding environment. Basically, exercise-induced neuroprotective mechanisms provide an example of the molecular adaptation of activated motoneurons.