The recurrence of certain cancers remains quite high due to either incomplete surgical removal of the primary tumor or the presence of small metastases that are invisible to the surgeon. Near infrared (NIR) fluorescence imaging might improve surgical outcomes by providing sensitive, specific, and real-time visualization of normal and diseased tissues if agents can be found that discriminate between normal and diseased tissue and define tumor margins. We have developed a new approach for revealing tumor borders by using NIR fluorescently labeled pH Low Insertion Peptide (pHLIP) and have created a computational program for the quantitative assessment of tumor boundaries. The approach is tested in vivo by co-localization of GFP-tumors and NIR emission from the fluorescently labeled pHLIP, and it is found that boundaries are accurately reported and that sub-millimeter masses can be detected.