In this study, a Langevin model is constructed by modifying a neuron–astrocyte coupled model that comprises a pyramidal neuron, an interneuron, and an astrocyte. This Langevin model considers random open-close transitions of calcium ion channels in the endoplasmic reticulum membrane of astrocytes. The effect of noise in the astrocytes on neuronal transmission is investigated numerically based on a random model under both normal and overexpression conditions of metabotropic glutamate receptors on astrocyte membranes. This study suggests that noise could change the firing patterns of two neurons during neuronal information transmission. Noise facilitates the occurrence of episodic spikes (ESs) in both neurons. However, the noise-induced ESs occur irregularly, compared with ESs in a deterministic model, and the change in regularity with noise exhibits the coherence- resonance phenomenon. Furthermore, synchronicity between noisy ESs in two neurons depends significantly on various parameters. ESs completely occur synchronously but irregularly in certain parameter regions, whereas ESs in other parameter values are antiphase synchronous. This study implies not only that the calcium dynamics in astrocytes could participate in neuronal transmission, but also that noise in astrocytes may be transferred to neurons and may affect synaptic transmission significantly.