The differential evolution (DE) is a simple and effective stochastic algorithm for the global optimization. The algorithm has only a few parameters. Nevertheless, setting these parameters to the appropriate values, when the particular optimization problem is solved, can take lot of time. Due to this fact several adaptive approaches have appeared recently. The aim of this work is to propose a new adaptive variant of DE. The new variant is experimentally compared with the state-of-the-art adaptive approaches in DE on the set of problems developed for the special session of CEC2005.