GnRH secretion pulsatility is required for adequate reproductive physiology. As in vitro models of GnRH neurones developed, information has been gained on cellular and molecular mechanisms governing neuronal activity at the single neurone level. However cellular and molecular mechanisms involved in pulsatile secretion from the whole network remain largely unknown. In vitro models of GnRH neurones network showed that GnRH pulses were correlated with calcium events synchronization. We showed that the repartition of calcium events followed a Poisson law; each GnRH neurone having its own law, and that there was no spatial coherence at the time of high synchronization events. Analysis of neuronal synchronization in the presence of the GnRH receptor antagonist, antide, showed the disappearance of high synchronization events without affecting GnRH pulsatile secretion. This result suggested that GnRH locally released caused synchronization of surrounding neurones. Therein GnRH secretion might be at the origin of the stochastic behaviour within the network. To better understand the behaviour of the GnRH neuronal network, we developed a mathematical model based on stochastic properties of GnRH vesicles trafficking. The core of the model is a three compartments model : 1. Factory (GnRH synthesis and packaging), 2. Stock pool and 3. Immediately releasable pool (IRP). The release of vesicles from IRP is described by a Poisson process of parameter lambaR; which depends on cytoplasmic free calcium concentration and indirectly on extracellular GnRH concentration; beyond a threshold values G, the expected value of released vesicles will be small and above this critical value GnRH will act as a autocrine/paracrine positive feedback. This model predict a built up of IRP when GnRH secretion is stopped. Decreasing electrical activity experimentally, strongly decreased GnRH pulsatile secretion, blocked calcium events synchronization and was associated with an increase GnRH immunoreactivity close to neurone membrane.