The university growth and the increase in university libraries, in addition to the reduction of available resources, make necessary performance evaluation systems, that allow managers to assess the quality of services provided to users and the use of resources. A performance measurement tool used in the performance evaluation of libraries (university and public) is Data Envelopment Analysis (DEA). The present study aims to develop a decision support approach that allows the resource reallocation between the libraries of the Integrated System of Libraries (SISBI) of Federal University of Ouro Preto (UFOP) using scientifically based criteria. For this, BCC classic and BCC with input zero-sum gains (DEA-GSZ BCC) models were used. In the DEA-GSZ BCC model, we chose to redistribute only the HOR input (which represents the number of hours worked in each library per day), since the redistribution of the other inputs related to the collections and buildings of the libraries is not justifiable. The results suggest an improvement in the performance of inefficient libraries, as well as an increase in the overall efficiency of the library system due to the more efficient use of resources.