Trajectories of solid radioactive tracers freely moving in a 3D gas-liquid-solid fluidized bed are analyzed from a data mining approach and using tools of symbolic dynamics. The trajectories are obtained by the radioactive particle tracking (RPT) technique, employing solids of different densities (glass beads and hydrophilized PVC particles), tap water as the fluidizing media and filtered air as the gas phase. Different gas velocities are examined to span the homogeneous and the heterogeneous flow regimes. The aim is to define parameters that can act as flow regime pointers. Hence, particular features of the particle motion are examined in detail to relate them to the underlying flow regime. In addition, by means of data symbolization, it is found that the recurrence of specific patterns of the tracer trajectories leads to higher frequencies of certain symbol sequences. Estimated frequencies are strongly influenced by the flow regime and may allow for regime identification. (c) 2005 Elsevier B.V. All rights reserved.