Key points L‐type calcium channels in the CNS exist as two subunit forming channels, Cav1.2 and Cav1.3, which are involved in short‐ and long‐term plasticity. We demonstrate that Cav1.3 but not Cav1.2 is essential for wind‐up. These results identify Cav1.3 as a key conductance responsible for short‐term sensitization in physiological pain transmission. We confirm the role of Cav1.2 in a model of long‐term plasticity associated with neuropathic pain. Up‐regulation of Cav1.2 and down‐regultation of Cav1.3 in neuropathic pain underlies the switch from physiology to pathology. Finally, the results of the present study reveal that therapeutic targeting molecular pathways involved in wind‐up may be not relevant in the treatment of neuropathy. Abstract Short‐term central sensitization to pain temporarily increases the responsiveness of nociceptive pathways after peripheral injury. In dorsal horn neurons (DHNs), short‐term sensitization can be monitored through the study of wind‐up. Wind‐up, a progressive increase in DHNs response following repetitive peripheral stimulations, depends on the post‐synaptic L‐type calcium channels. In the dorsal horn of the spinal cord, two L‐type calcium channels are present, Cav1.2 and Cav1.3, each displaying specific kinetics and spatial distribution. In the present study, we used a mathematical model of DHNs in which we integrated the specific patterns of expression of each Cav subunits. This mathematical approach reveals that Cav1.3 is necessary for the onset of wind‐up, whereas Cav1.2 is not and that synaptically triggered wind‐up requires NMDA receptor activation. We then switched to a biological preparation in which we knocked down Cav subunits and confirmed the prominent role of Cav1.3 in both naive and spinal nerve ligation model of neuropathy (SNL). Interestingly, although a clear mechanical allodynia dependent on Cav1.2 expression was observed after SNL, the amplitude of wind‐up was decreased. These results were confirmed with our model when adapting Cav1.3 conductance to the changes observed after SNL. Finally, our mathematical approach predicts that, although wind‐up amplitude is decreased in SNL, plateau potentials are not altered, suggesting that plateau and wind‐up are not fully equivalent. Wind‐up and long‐term hyperexcitability of DHNs are differentially controlled by Cav1.2 and Cav1.3, therefore confirming that short‐ and long‐term sensitization are two different phenomena triggered by distinct mechanisms.