This paper presents a new method to control the delay performance for high rate heterogeneous real-time traffic flows based on a novel traffic control algorithm which is a generalization of traditional (\sigma, p) regulator. Our new control algorithm operates like the traditional regulator under the normal loading situation, but provides more regulation for the high rate (heavy load condition) of the traffic. For a set of heterogenous real-time traffic flows R, we can show that Dr(R) \le D(R) where D_r(R) and D(R) are the worst-case delay bounds with our new control algorithm and that with (\sigma, p) regulator, respectively. More specifically, we develop a set of formula that can be used to set the parameters in our new traffic controller so that the worst case delay bound is minimized by streaming the traffic flow. We can prove that there exists a minimum (average) input rate p* such that D_r(R) = D(R) for p \le p* and D_r(R) < D(R) for p > p*. Using the extended regulator can effectively control the delay when the average heterogeneous traffic rate is high. The issues are particularly useful for Integrated Services where a flow may over claim its share of resource and for Differentiated Services where a class of traffic flows may posses very high rates.