Dynamic real-time systems often operate in a continuously changing environment, causing workload of the system to fluctuate. An initial resource allocation for these systems should be robust with respect to the variation in workload. Using the amount of additional workload that an allocation can accommodate as a measure of robustness, we develop an iterative integer programming approach, called IIP, to determine a robust resource allocation. IIP guarantees to produce an allocation with the measure of robustness that falls within d from the optimal value, where d is a user provided parameter for the IIP algorithm. In addition, trade-off between the quality of the resulting allocation and the execution time of IIP can be achieved by adjusting the parameter d.