Background The detection of antibodies to myelin oligodendrocyte glycoprotein (MOG) is fundamental for the identification of MOG antibody-associated disorders (MOGAD), and the differential diagnosis of acquired demyelinating syndromes of the CNS, among which multiple sclerosis (MS). We compared the diagnostic performance of four cell-based assays (CBAs) for their detection. Methods Consecutive sera from 204 patients with ‘possible MOGAD’ (55), MS (112), and other neurological disorders (OND, 37) were tested for MOG-IgG with a live-CBA with anti-heavy-and-light chain secondary-antibody (LCBA-IgG H+L ), and a live-CBA for IgG 1 (LCBA-IgG 1 ). A subgroup of 71 patients was additionally tested with a live-CBA with anti-Fcγ secondary-antibody (LCBA-IgG Fcγ ), and a commercial fixed-CBA with anti-Fcγ secondary-antibody (FCBA-IgG Fcγ ) . Results Fifty-seven/204 patients (27.9%) were MOG-IgG-positive. Sensitivity was 89.1% (CI:77.8–95.9) and specificity 93.3% (CI:88.0–96.7) for LCBA-IgG H+L , and 74.6% (CI:61.0–85.3) and 100% (CI:97.6–100) for LCBA-IgG 1 . Eighteen of 57 (31%) samples showed discrepant results (all negative on LCBA-IgG 1 ); of these, three with ‘possible MOGAD’ showed high-titer MOG-IgG (≥ 1:640), and positivity for MOG-IgG 2 , whereas 15/18 had low-titer MOG-IgG (1:160/1:320) and mixed diagnoses (5 ‘possible MOGAD’, 6 MS, 4 OND). In the subgroup analysis, sensitivity was 92.3% (CI:79.1–98.4) and specificity 97.0% (CI:83.8–99.9) for LCBA-IgG Fcγ , and 87.2% (CI:72.6–95.7) and 97.0% (CI:83.8–99.9) for FCBA-IgG Fcγ . Conclusions LCBA-IgG 1 showed the highest specificity but can miss MOG-IgG 2 reactivities, whose meaning warrants further investigations. Titration of samples tested with LCBA-IgG H+L / IgG Fcγ is important for meaningful interpretation of the results. In the subgroup analysis, LCBA-IgG Fcγ yielded the highest accuracy, and FCBA-IgG Fcγ good specificity, but it was at risk of false-negative results.