Increasing experience indicates that anti-Ma2-associated encephalitis differs from classical paraneoplastic limbic or brainstem encephalitis, and therefore may be unrecognized. To facilitate its diagnosis we report a comprehensive clinical analysis of 38 patients with anti-Ma2 encephalitis. Thirty-four (89%) patients presented with isolated or combined limbic, diencephalic or brainstem dysfunction, and four with other syndromes. Considering the clinical and MRI follow-up, 95% of the patients developed limbic, diencephalic or brainstem encephalopathy. Only 26% had classical limbic encephalitis. Excessive daytime sleepiness affected 32% of the patients, sometimes with narcolepsy-cataplexy and low CSF hypocretin. Additional hormonal or MRI abnormalities indicated diencephalic-hypothalamic involvement in 34% of the patients. Eye movement abnormalities were prominent in 92% of the patients with brainstem dysfunction, but those with additional limbic or diencephalic deficits were most affected; 60% of these patients had vertical gaze paresis that sometimes evolved to total external ophthalmoplegia. Three patients developed atypical parkinsonism, and two a severe hypokinetic syndrome with a tendency to eye closure and dramatic reduction of verbal output. Neurological symptoms preceded the tumour diagnosis in 62% of the patients. Brain MRI abnormalities were present in 74% of all patients and 89% of those with limbic or diencephalic dysfunction. Among the 34 patients with cancer, 53% had testicular germ-cell tumours. Two patients without evidence of cancer had testicular microcalcification and one cryptorchidism, risk factors for testicular germ-cell tumours. After neurological syndrome development, 17 of 33 patients received oncological treatment (nine also immunotherapy), 10 immunotherapy alone, and six no treatment. Overall, 33% of the patients had neurological improvement, three with complete recovery; 21% had long-term stabilization, and 46% deteriorated. Features significantly associated with improvement or stabilization included, male gender, age <45 years, testicular tumour with complete response to treatment, absence of anti-Ma1 antibodies and limited CNS involvement. Immunosuppression was not found to be associated with improvement but was clearly effective in some patients. Fifteen patients (10 women, five men) had additional antibodies to Ma1. These patients were more likely to have tumours other than testicular cancer and to develop ataxia, and had a worse prognosis than patients with only anti-Ma2 antibodies (two women, 21 men); 67% of deceased patients had anti-Ma1 antibodies. Anti-Ma2 encephalitis (with or without anti-Ma1 antibodies) should be suspected in patients with limbic, diencephalic or brainstem dysfunction, MRI abnormalities in these regions, and inflammatory changes in the CSF. In young male patients, the primary tumour is usually in the testis, in other patients the leading neoplasm is lung cancer.