Abstract-3GPP specifications allow to change the physical resources allocated to the users dynamically to optimize the radio resource utilization in UMTS. Network Operators can make use of this optimization feature by properly adjusting dynamic bearer type switching in their networks. In this paper we show an analysis of the dynamic bearer type switching based on measurements in a live UMTS network in the city center of Vienna, Austria. Particularly, we present the usage probability and the runlength distribution of the three different radio access bearers which are used for transmission over the UMTS DCH (Dedicated Channel) packet switched (PS) domain in DL (Down Link), namely 384kbit/s, 128kbit/s and 64kbit/s bearer. For the performance evaluation of applications, system level simulators are used, where the lower layers are usually represented via stochastic models. Following the measured statistics we develop a model capable of properly describing the dynamic switching between the mentioned bearers. Furthermore, we show that a four-state Markov model with its geometrically distributed runlengths does not provide sufficient accuracy. We achieve an enhancement by using a four-state renewal process with Weibull distributed runlengths as presented in this document.