This paper presents a set of mechanisms for congestion control of data traffic in high speed wide area networks (HSWANs) along with preliminary performance results. The model of the network assumes reservation of resources based on average requirements. The mechanisms address (a) the different network time constants (short term and medium-term), (b) admission control that allows controlled variance of traffic as a function of medium-term congestion, and (c) prioritized scheduling which is based on a new fairness criterion. This latter criterion is perceived as the appropriate fairness measure for HSWANs. Preliminary performance studies show that the queue length statistics at switching nodes (mean, variance and max) are approximately proportional to the end-point 'time window' size. Further, * when network utilization approaches unity, the time window mechanism can protect the network from buffer overruns and excessive queueing delays, and * when network utilization level is smaller, the time window may be increased to allow a controlled amount of variance that attempts to simultaneously meet the performance goals of the end-user and that of the network. The prioritized scheduling algorithms proposed and studied in this paper are a generalization of the Virtual Clock algorithm [Zhang 1989]. The study here investigates * necessary and sufficient conditions for accomplishing desired fairness, * simulation and (limited analytical results for expected waiting times, * ability to protect against misbehaving users, and * relationship between end-point admission control (Time-Window) and internal scheduling ('Pulse' and Virtual Clock) at the switch.