We hereby apply the approach of photoacoustic chemical imaging, performing an in vivo chemical analysis that is spatially resolved (200 μm) and in real time, to predict a given tumor’s response to therapy. Using triple negative breast cancer as a model, we took photoacoustic images of tumors’ oxygen distributions in patient-derived xenografts (PDXs) in mice using biocompatible, oxygen-sensitive tumor-targeted chemical contrast nanoelements (nanosonophores), which function as contrast agents for photoacoustic imaging. Following radiation therapy, we established a quantitatively significant correlation between the spatial distribution of the initial oxygen levels in the tumor and its spatial distribution of the therapy’s efficacy: the lower the local oxygen, the lower the local radiation therapy efficacy. We thus provide a simple, noninvasive, and inexpensive method to both predict the efficacy of radiation therapy for a given tumor and identify treatment-resistant regions within the tumor’s microenvironment.