Conventionaldata envelopment analysis (DEA) models in the performance evaluation are based on black box thought, so that inputs in the boxes (decision making units) are converted into outputs.One of the drawbacks of these models is the neglection of internal structure, intermediate products or linking activities; also the actual transforming process generally isn’t modeled explicitly. According to the multi-stage structure of the banking industry; in this research, after pointing out black box (aggregation) and separation approaches, needs for inclusion of internal process of the decision making units (DMUs) expressed and a slacks-based network data envelopment analysis model is introduced and used in performance evaluation of the banking industry. The major merit of slacks-based measure is its ability to provide suitable efficiency measures, especially for weakly efficient units. Among the findings of thisresearch is that in the structures with linking and networking effects, using the black box and separation approaches, don’t produce actual assessment of performance and should be used network models.