This article presents an algorithm for the synthesis of heat exchanger networks (HENs) using randomization as an effective tool. It optimizes the total cost of the network. The method proposed here is suitable for finding the optimal solution with stream-splitting and merging. The present approach provides significant advantage of randomization over other existing optimization techniques for obtaining the optimal solution. We have studied three benchmark problems already published in the literature to demonstrate how better solutions were undetected by the earlier approaches. The salient feature of the proposed algorithm is that it provides a variety of possible networks which are close to the optimal network. Another important aspect is that the algorithm is quite fast. For small-and medium-size problems, the technique proposed in this article will prove to be very effective for the design of heat exchanger networks. (C) 2009 Elsevier Ltd. All rights reserved.