Abstract： To investigate the risk factors of poor prognosis and recurrence in patients with anti-NMDAR encephalitis. A single center, observational cohort study was used to retrospectively analyze 44 patients with anti NMDAR encephalitis hospitalized in the Department of Neurology of Beijing Tong Ren Hospital from January 2014 to October 2020. The results showed that the interval from onset to immunotherapy in the poor prognosis group was significantly longer than that in the good prognosis group ( t=2.045, P=0.047), and the course of disease in the poor prognosis group was significantly longer than that in the good prognosis group ( t=4.127, P=0.000 2). The number of patients with clinical manifestations of dyskinesia was significantly increased (Fisher exact test: P=0.014). The patients with abnormal brain MRI in the poor prognosis group were significantly more than those in the good prognosis group (Fisher exact test: P=0.017), and the patients with slow wave>50% in the poor prognosis group were significantly more than those with slow wave <50% (Fisher exact test: P<0.001). Patients with the first onset of immunotherapy time <3 months, long course of disease, high intracranial pressure, and high cerebrospinal fluid protein are prone to relapse. Bivariate logistic regression analysis showed that patients with dyskinesia, abnormal brain MRI, and slow wave EEG more than 50% were risk factors for poor prognosis ( OR values were 4.687, 4.978, and 24.500, respectively; P values were 0.018, 0.016, and 0.000, respectively). The time of first-line immunotherapy for the first onset<3 months was the risk factor for recurrence ( OR 17.231, P=0.010). In conclusion, dyskinesia, abnormal brain MRI and slow wave of EEG more than 50% may be the risk factors for poor prognosis of patients. The duration of immunotherapy less than 3 months after the first onset might be the risk factor for recurrence.