Experiments were performed on neonatal mice to analyze why, in vitro, the respiratory rhythm generator (RRG) was silent and how it could be activated. We demonstrated that in vitro, the RRG in intact brainstems is silenced by a powerful inhibition arising from the pontine A5 neurons through medullary a2 adrenoceptors and that in vivo, nasal trigeminal inputs facilitate the RRG since nasal continuous positive airway pressure increases the breathing frequency whereas nasal occlusion and nasal afferent anesthesia depress it. Since nasal trigeminal afferents project to the A5 nuclei, we applied single trains of negative electric shocks to the trigeminal nerve in inactive ponto-medullary preparations. They induced rhythmic phrenic bursts during the stimulation and for 2-3 minutes afterwards whereas repetitive trains produced on-going rhythmic activity up to the end of the experiments. Electrolytic lesion or pharmacological inactivation of the ipsilateral A5 neurons altered both the phrenic burst frequency and occurrence after the stimulation. Extracellular unitary recordings and trans-neuronal tracing experiments with the rabies virus show that the medullary lateral reticular area contains respiratory-modulated neurons, not necessary for respiratory rhythmogenesis, but which may provide an excitatory pathway from the trigeminal inputs to the RRG since their electrolytic lesion suppresses any phrenic activity induced by the trigeminal nerve stimulation. The results lead to the hypothesis that the trigeminal afferents in the mouse neonate involve at least two pathways to activate the RRG, one which may act through the medullary lateral reticular area and one which releases the A5 inhibition received by the RRG.