This paper presents the results of angle-delay measurements in physically nonstationary radio channels. The multidimensional estimation data are obtained using a recently developed 3-D high-resolution channel sounder. For a better analysis, a hierarchical whole clustering method is presented that organizes the multidimensional estimation data into clusters. Furthermore, estimation results are compared with results obtained from a 3-D deterministic propagation prediction tool. The clustering method proposed here was successfully applied to multi-dimensional estimation data. The results show that it is vital to cluster these data in order to be able to correctly interpret the measurement results. Compared to the measurements, a large number of multipath clusters are missing in the predictions. Furthermore, the multipath clusters obtained from the measurement data tend to have larger time and angular spreads than predicted.