In the last few years, a rapidly growing number of autoantibodies targeting neuronal cell-surface antigens have been identified in patients presenting with neurological symptoms. Targeted antigens include ionotropic receptors such as N-methyl-D-aspartate receptor or the α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor, metabotropic receptors such as mGluR1 and mGluR5, and other synaptic proteins, some of them belonging to the voltage-gated potassium channel complex. Importantly, the cell-surface location of these antigens makes them vulnerable to direct antibody-mediated modulation. Some of these autoantibodies, generally targeting ionotropic channels or their partner proteins, define clinical syndromes resembling models of pharmacological or genetic disruption of the corresponding antigen, suggesting a direct pathogenic role of the associated autoantibodies. Moreover, the associated neurological symptoms are usually immunotherapy-responsive, further arguing for a pathogenic effect of the antibodies. Some studies have shown that some patients’ antibodies may have structural and functional in vitro effects on the targeted antigens. Definite proof of the pathogenicity of these autoantibodies has been obtained for just a few through passive transfer experiments in animal models. In this review we present existing and converging evidence suggesting a pathogenic role of some autoantibodies directed against neuronal cell-surface antigens observed in patients with central nervous system disorders. We describe the main clinical symptoms characterizing the patients and discuss conflicting arguments regarding the pathogenicity of these antibodies.