Aquaporin-4 (AQP4) antibody sero-positivity is critically important in neuromyelitis optica (NMO). However, the sensitivity of different assays is highly variable. Repeating detection with a highly sensitive assay in a large population is necessary in the case of so-called negative NMO. Retrospective analysis where AQP4 antibodies were detected by commercial cell-based assay (CBA), in-house M23-CBA and in-house M1-CBA. Of the 1011 serum samples, 206 (20.4%) were sero-positive by primary commercial CBA. In the retest, all 206 participants positive by primary commercial CBA also yielded positive results by in-house M23-CBA and the second commercial CBA again, but only 124 positive in in-house M1-CBA. Among the 805 participants negative by primary commercial CBA, 71 participants were positive for in-house M23-CBA, of which 20 participants were positive for the second commercial CBA, and none were positive by in-house M1-CBA. Of the 171 cerebral spinal fluid samples, 75 (43.9%) were positive by primary commercial CBA. All 75 participants positive by primary commercial CBA also yielded positive results by in-house M23-CBA and the second commercial CBA. Forty-nine (65.3%) of these 75 participants were positive by in-house M1-CBA. Among the 96 participants negative by primary commercial CBA, 15 participants were positive for in-house M23-CBA and none were positive by in-house M1-CBA and the second commercial CBA. Different AQP4 isoforms in CBA result in different detection effects, and in-house M23-CBA is the most sensitive method. Some AQP4 antibody-negative NMO may be subject to diagnostic uncertainty due to limitations of the assays.