This focus of this work is the development of resource management schemes for communication networks. Model based approaches using effective and decoupling bandwidth constraints that guarantee both individual and aggregate quality of service requirements are investigated. The emphasis is on practical issues, as well as on understanding the accuracy of these approximations at the network level. Further study of these concepts should elucidate the impact of flow control on network capacity. Since, in practice, traffic statistics and demand may be unknown or time varying, methods for on-line estimation and control are also investigated. For example, a self-tuning rate control throttle is proposed which could in principle minimize a traffic stream's burstiness subject to a loss or delay constraint by monitoring the arrival process. This is a first step towards intelligent networks that monitor, learn, and adapt to the traffic statistics and demands. The final goal of this research is to develop comparison methods based on the "robustness" of a performance guarantee for a given traffic stream.