Spouted beds are commonly simulated through the Computational Fluid Dynamics - Discrete Element Method approach. The choice of the drag model is still a matter of debate, as they feature peculiar operative conditions. In this work, we simulated two spouted beds containing Geldart-D particles. We tested seven drag models: three are classic models, while four are developed through advanced computational techniques. The results indicate that the key variable is the ratio between the operative and the minimum spouting gas velocity (u/u(ms)). At u = u(ms) only the Gidaspow model can always predict fluidisation, but at low u/u(ms) values the Beetstra model is the best compromise. For higher values, the Rong and Di Felice models behave better, while the others overestimate the particles' velocity. These results can be useful to identify the best performing model and show there is a need for more appropriate models for spouted beds. (C) 2019 Elsevier B.V. All rights reserved.