Objective There are now a large number of requests for N-methyl-d-aspartate receptor autoantibody (NMDAR-Ab) tests, and it is important to assess the clinical relevance of all results, particularly when they are reported as ‘Low Positive’.

Methods The clinical data of 56 patients found Positive or Low Positive by the Oxford live cell-based assay were reviewed. An autoimmune basis for the condition was assigned as ‘Definite’, ‘Possible’ or ‘Unlikely’. The number of core features (encephalopathy, psychiatric, cognitive, epileptic, extrapyramidal and inflammatory cerebrospinal fluid (CSF)) was tabulated.

Results Twenty-five (44.6%) patients had a Definite NMDAR-Ab encephalitis (eight ovarian teratomas, one Hodgkin's lymphoma), 18 (32.1%) a Possible NMDAR-Ab encephalitis and 13 (23.2%) an Unlikely autoimmune syndrome. Serum NMDAR-Ab levels were higher in patients with tumours. Positive NMDAR-Abs were found not only in patients with three or more core features and a Definite syndrome, but also in five patients classified as Possible. Conversely, Low Positive NMDAR-Abs were present in 7 Definite cases as well as in 13 Possible cases. Unlikely patients had mainly Low Positive antibodies and fewer core features. CSF NMDAR-Abs, only available in 11 pairs and at varying time points, broadly related to serum levels and were Positive in 3/3 patients with tumours but in only 2/5 Definite patients, and none of the Possible or Unlikely cases.

Interpretation Using live cell-based assays, Positive and Low Positive antibodies can be of clinical significance. The number of core clinical features should help to select those patients in whom an immunotherapy intervention might be considered, irrespective of the antibody level.