Abstract: This study compares the results of two commonly used computational fluid dynamics methods for fluidized bed simulation of smooth type-A monodisperse particles. One approach is based on the continuum assumption called the two-fluid-method (TFM), and one discrete particle method (DEM) based on a soft-sphere collision algorithm. Both DEM and TFM are coarsened by either lumping several particles in a parcel (CDEM) and/or by increasing the computational mesh size to achieve faster simulation speed. Simulations using both methods were conducted in a 3D periodic cylindrical riser domain in the core-annulus flow regime. Four different levels of coarseness were applied for each method with the finest simulations utilizing more than 200 M particles and 2.9 M computational cells. The finest DEM simulation is assumed to yield the most accurate results as attested in the literature. An error analysis shows that it is possible for CDEM to yield faster and more accurate results than TFM, although TFM is generally faster at the same coarseness level. For more complex cases involving particles with statical properties, such as size and density distributions, CDEM is anticipated to be a better choice for fluidized bed simulations.