The introduction of third generation (3G) public wireless network infrastructures, such as the Universal Mobile Telecommunications System (UMTS), enable the development of innovative mobile services. For example, deploying m-health services which embed tele-monitoring and tele-treatment services become feasible with the role-out of 3G networks. These services allow healthcare professionals to monitor a mobile patient's vital signs and provide feedback to this patient anywhere and any time. The performance of m-health services perceived by end-users depends on the serviceableness of 3G networks to support these services. Hence, the performance of 3G networks is a critical factor for successful development of m-health services. In this paper, we present a methodology for measurements-based performance assessment of 3G networks that aim to support m-health services. This methodology has been applied to evaluate end-user perceived service performance, in relation to the performance (i.e. serviceableness) of a 3G network. In addition, we analyse the measurements with the purpose to improve the end-to-end delay characteristics of the telemonitoring service as well as optimize the (derived) goodput behaviour of this 3G network. Our results show that the goodput behaviour is asymmetric and depends on a bearer assignment policy of the network. Based on our results we provide guidelines for the design of application protocols for m-health services and how these protocols deal with the changing performance behaviour of 3G networks.