OBJECTIVE Methotrexate requires therapeutic drug monitoring in oncology because of narrow therapeutic index, especially the metabolite 7-hydroxymethotrexate exhibits nephrotoxicity. The goal of this study was to evaluate different assays and their impact on clinical decisions. METHODS Following routine measurement with Abbott TDxFLx® assay (MTX-TDX), 62 samples were analysed on Architect®i1000 (MTX-ARCHI), Xpand® (ARK/XPND), Indiko® (ARK/INDI), and HPLC (MTX-HPLC) as the reference method. The influence of 7-hydroxymethotrexate was explored on ARK reagent to document the cause of the observed bias. ROC curves were built to study the impact of the method on the discharge thresholds for the patients at three levels. RESULTS Total imprecision was below 2.60% for the methotrexate-ARCHI and close to 10% for both ARK assays for plasma pools. The correlation coefficients were 0.93, 0.93, 0.89 and 0.95, the Bland-Altman difference plot revealed a bias of 0.075, 0.037, 0.049 and -0.002, and the number of results exceeding the TE criteria of 0.1 µM was 17 (27%), 13 (21%), 15 (24%) and 15 (24%) for MTX-TDX, ARK/INDI, ARK/XPND and MTX-ARCHI, respectively. Cross reactivity with 7-hydroxymethotrexate was between 1 and 9%. Overestimation of methotrexate concentration was between -4% and +32%. The most robust clinical level was found to be the highest level (0.2 µM) with ROC curves. CONCLUSIONS The authors found the best results for imprecision with chemiluminescent microparticle immunoassay method on methotrexate-ARCHI, with bias below to the RICOS recommendations and best correlation to the reference method. Impact on the threshold values for clinical decision need to be clearly exposed to clinicians.