Medical imaging has made significant contributions to the characterization of malignant tumors. In many cases, however, maps from multiple modalities may be required for more complete tumor mapping. In this manuscript we propose an objective method for combining multiple imaging datasets with the goal of characterizing malignant tumors. We refer to the proposed technique as the percent overlap method (POM). To demonstrate the power and flexibility of the POM analysis, we present four patients with recurrent glioblastoma multiforme. Each patient had multiple magnetic resonance imaging procedures resulting in seven different parameter maps. Chemical shift imaging was used to provide three metabolite ratio maps (Cho:NAA, Cho:Cre, Lac:Cre). A perfusion scan provided regional cerebral blood volume and permeability maps. Diffusion and carbogen-based hypoxia mapping data were also acquired. Composite maps were formed for each patient using POM, then were compared to results from the ISODATA clustering technique. The POM maps of likely recurrent tumor regions were found to be consistent with the ISODATA clustering method. This manuscript presents an objective method for combining parameters from multiple physiologic imaging techniques into a single composite map. The accuracy of the map depends strongly on the sensitivity of the chosen imaging parameters to the disease process at the time of image acquisition. Further validation of this method may be achieved by correlation with histological data.