Objectives Measurement of the serum levels of vitamin B12 (VB12) is key for evaluating VB12 deficiency-dependent anemia. Immunoassay, the major method for determining VB12, tends to give false-normal results because of the presence of anti-intrinsic factor (IF-Ab) or other factors such as heterophilic antibodies et al. This study aimed to develop a liquid chromatography tandem mass spectrometry (LC-MS/MS) method that is helpful for distinguish false normal VB12 results measured by the immunoassay. Methods Different forms of VB12 were derivatized into CN-B12, which was collected through solid-phase extraction and analyzed via LC-MS/MS. 236 serum samples were measured both by LC-MS/MS and immunoassay, results were compared, and the IF-Ab effect was evaluated. Results The LC-MS/MS assay afforded a linear slope from 20 to 4,000 pmol/L for CN-B12. OH-VB12, methyl-VB12, and CoA-VB12 showed recovery within 89.3-109.5%. The intra-assay CV of VB12 was 2.6-4.1%, whereas the total CV was 9.3-9.8%. Passing-Bablok regression between LC-MS/MS and immunoassay results showed that the slope was 1.085 and the intercept was -15.691. The Bland-Altman plot showed that the mean difference and difference% were -34.6 pmol/L and 0.3%, respectively. Inter-rater agreement analysis showed that the linear weighted kappa value was 0.885, implying good agreement between the two methods. However, two samples were falsely elevated and one sample was falsely normal in the immunoassay compared with LC-MS/MS. The LC-MS/MS method helped in the distinction of false-normal VB12 results shown by the immunoassay. Conclusions The VB12 LC-MS/MS method can be used as an arbiter of clinically discordant immunoassay results.