Abstract: FY-3D MERSI/NDVI is a critical operational product used in many studies (ecosystem monitoring, climate change, agriculture drought, etc.), and it is essential to obtain a comprehensive assessment of this product's quality. In this paper, the first assessment results of global MERSI/NDVI for the period from May 2019 to December 2020 are reported using Terra MODIS/NDVI as the reference. Quantitative measures of APU (Accuracy, Precision and Uncertainty) are calculated and the variations associated with different factors (seasons, land cover types and NDVI values) are analyzed. Results indicate that generally the two products share a high similarity concerning the spatial pattern and temporal profiles features. The dynamic range of MERSI/NDVI is slightly narrower because it overestimates NDVI for barren land and underestimates NDVI for dense vegetation. Sensitivity analysis indicates that the overestimation is mainly attributed to overestimation of NIR reflectance, whereas the underestimation is mainly attributed to overestimation of red reflectance. Phonological features conveyed by the two NDVI products are consistent, but there are slightly noisier fluctuations in MERSI/NDVI time series probably caused by cloud contamination during growing season. Over the 20 months period checked in this study, the global mean of the Accuracy value ranges within −0.02—0, and the global mean of the Precision and Uncertainty values generally range within 0.06—0.08. With respect to the spatial pattern, APU values are the highest in forests, moderate in grassland/shrubland, and lowest in desert. Linear regression models with MODIS/NDVI as independent variables and MERSI/NDVI as dependent variables achieve high accuracies (R²: 0.91—0.95, RMSE: 0.048—0.068), confirming that it is feasible to build a compatible long-term NDVI dataset using both products. This study is the first cross-sensor comparison study using almost all of the global MERSI/NDVI data available since the operational application of the FY-3D satellite. Overall, the MERSI/NDVI data are of very high quality and can be effectively used for deriving vegetation phonological and greenness information. Its performance on the global scale will be monitored continually.