The detection of compartments and dead zones as well as the estimation of the mixing efficiency in stirred tanks are of vital interest for a variety of biochemical and chemical processes. Here, numerically derived time-dependent 3D fluid velocity fields of a stirred tank reactor are computed using the Lattice Boltzmann Method. Mixing in the stirred tank reactor is analysed by means of Lagrangian Coherent Structures which allow to unravel the mixing states of complex flows. This Lagrangian analysis is achieved by computing Finite Time Lyapunov Exponents and applying recent trajectory-based network methods on the three-dimensional flow. The results reveal a zone of low interaction in the upper region of the stirred tank reactor and five additional transient compartments. The trajectory-based network analysis detects low cross-mixing of fluid parcels between different compartments but a very high mixing of fluid parcels inside of each compartment. This high interaction is also found in an analysis of the Finite Time Lyapunov Mixing Intensity. Time-averaging of the fluid velocity field prior to the Lagrangian analysis is considered to extract the most influential Lagrangian Coherent Structures.