Aggregation of data flows has two major advantages. One is the reduction of state complexity within the network, the other is the saving of resources by statistical multiplexing between the aggregated flows. In this paper, we show a simple, yet effective scheme to aggregate real-time flows which require a statistical guarantee on experienced loss and a deterministic guarantee on the maximum delay on an individual basis. The focus of our aggregation scheme is on the reduction of state complexity. Therefore, we try to maximize the number of flows to be aggregated by the consideration of heterogeneous flows at the cost of maximally saving resources which would require homogenous flows to be aggregated. Our approach is to first gain insight on the buffer occupancy distribution of a single flow. In practice, the buffer occupancy distribution function of a real-time flow can be considered as monotonic decreasing. We show that the uniform distribution, which is analytically very tractable, is always more pessimistic than a monotonic decreasing distribution. This allows us to aggregate heterogeneous flows by taking the uniform distribution as a worst-case bound for the individual flows’ buffer distributions and exploiting its statistical properties to save buffer resources by statistical multiplexing between the individual flows of the aggregate. Finally, we discuss at which rate such an aggregate of heterogeneous flows has to be served while maintaining the statistical guarantees given to individual flows.