ObjectivesThe purpose of this study was to investigate the association of neutrophil percentage-to-albumin ratio (NPAR) with the severity at admission and discharge (short-term prognosis) in patients with anti-N-methyl-D-aspartic acid receptor (NMDAR) encephalitis.MethodsMultivariable logistic regression models such as NPAR were constructed based on univariable regression results. Receiver operating characteristic (ROC) curves, nomograms, and concordance index (c-index) were used to evaluate the efficacy of the models in assessing disease severity at admission and predicting short-term prognosis, validated by bootstrap, Hosmer–Lemeshow goodness-of-fit test, calibration curves, and decision curve analysis.ResultsA total of 181 patients with anti-NMDAR encephalitis diagnosed at the First Affiliated Hospital of Zhengzhou University were included. The results showed that NPAR had good sensitivity and specificity in assessing disease severity at admission and predicting short-term prognosis. The multivariable logistic regression models based on NPAR and other influencing factors had good discrimination, consistency, accuracy, calibration ability, applicability, and validity in assessing the severity at admission and predicting short-term prognosis.ConclusionNPAR has good clinical value in assessing disease severity at admission and predicting short-term prognosis of patients with anti-NMDAR encephalitis.