Paraneoplastic cerebellar degeneration (PCD) is a rare immune-mediated disease that develops mostly in the setting of neoplasia and offers a unique prospect to explore the interplay between tumor immunity and autoimmunity. In PCD, the deleterious adaptive immune response targets self-antigens aberrantly expressed by tumor cells, mostly gynecological cancers, and physiologically expressed by the Purkinje neurons of the cerebellum. Highly specific anti-neuronal antibodies in the serum and cerebrospinal fluid represent key diagnostic biomarkers of PCD. Some anti-neuronal antibodies such as anti-Yo autoantibodies (recognizing the CDR2/CDR2L proteins) are only associated with PCD. Other anti-neuronal antibodies, such as anti-Hu, anti-Ri, and anti-Ma2, are detected in patients with PCD or other types of paraneoplastic neurological manifestations. Importantly, these autoantibodies cannot transfer disease and evidence for a pathogenic role of autoreactive T cells is accumulating. However, the precise mechanisms responsible for disruption of self-tolerance to neuronal self-antigens in the cancer setting and the pathways involved in pathogenesis within the cerebellum remain to be fully deciphered. Although the occurrence of PCD is rare, the risk for such severe complication may increase with wider use of cancer immunotherapy, notably immune checkpoint blockade. Here, we review recent literature pertaining to the pathophysiology of PCD and propose an immune scheme underlying this disabling disease. Additionally, based on observations from patients' samples and on the pre-clinical model we recently developed, we discuss potential therapeutic strategies that could blunt this cerebellum-specific autoimmune disease.