This paper presents both theoretically and experimentally the advantages of using R-trees in content-based region query with spatial bounds. This advantage consists in the minimizing of the execution time. This process supposes that the user gives a rectangle by its coordinates in order to retrieve the images that are within these spatial bounds. The color regions of the images together with the information like color, area, the coordinates of the center of mass and the coordinates of the minimum bounding rectangle (MBR) have been already detected and stored in the database by using color set back-projection algorithm. There are presented a series of experiments in which the execution time of content-based region query with spatial bounds is measured in two cases: sequence query and the query using a R-tree that indexes the MBRs which frame the regions in the database.