Abstract: The transient behavior of rising bubbles plays a critical role on the performance of fluidized bed reactors, but predicting bubble dynamics is difficult. CFD has been shown to be capable of reproducing bubbling phenomena, but data interpretation and visualization is challenging. In this study, a 3-D detection and tracking algorithm, called face-masking, is developed and validated by numerical simulations of lab-scale and pilot-scale gas–solid fluidized beds. This algorithm identifies discrete bubbles using the instantaneous whole-field void fraction data. Individual bubbles are characterized in detail, including size, shape and location. The algorithm tracks bubbles across successive time frames and computes axial and lateral bubble velocities. Bubble dynamics predicted by the face-masking algorithm are validated against four different published experimental measurements. The face-masking algorithm provides a new tool for post-processing large-scale three-dimensional fluidizedbed simulation data. The bubble surfaces found by this algorithm will enable computation of normal fluxes through the surface. This algorithm can also be applicable in other areas of multiphase flows where characterization of bubbles, droplets, and clusters is necessary.