Grid infrastructures are distributed systems composed by heterogeneous resources with different geographical locations. Despite the advantages of such paradigm, it presents problems related to re- sources selection and availability.

The aim of this contribution is to provide a self-adaptive strategy for selecting grid resources efficiently during applications execution. Furthermore, the Scatter Search algorithm is applied during the selection process. The proposed model has been tested in a real grid infrastructure, achieving an execution time reduction and an improvement of the successfully completed tasks rate.