In this paper, a new random search technique which facilitates the determination of the global minimum, is presented. This method, called Adaptive Random Search Technique (ARSET), is experimented on test problems, and successful results are obtained. ARSET algorithm, outcome of which is observed to be relatively better, is also compared with other methods. In addition, applicability of the algorithm on artificial neural network training is tested with XOR problem. (c) 2005 Elsevier Inc. All rights reserved.