Fluorescently-quenched probes that are specifically activated in the cancer microenvironment have great potential application for diagnosis, early detection and surgical guidance. These probes are often designed to target specific enzymes associated with disease by direct optimization using single purified targets. However, this can result in painstaking chemistry efforts to produce a probe with suboptimal performance when applied in vivo. We describe here an alternate, unbiased activity-profiling approach in which whole tissue extracts are used to directly identify optimal peptide sequences for probe design. Screening of mouse mammary tumor extracts with a hybrid combinatorial substrate library (HyCoSuL) identified a combination of natural and non-natural amino acid residues that could be used to generate highly efficient tumorspecific fluorescently quenched substrate probes. The most effective probe is significantly brighter than any of our previously reported tumor imaging probes designed for specific proteases and robustly discriminates tumor tissue from adjacent healthy tissue in a mouse model of cancer.Importantly, although the probes were developed by screening mouse mammary tumor tissues, they are able to effectively distinguish human ductal carcinomas from normal breast tissue with similar reactivity profiles to those observed in mouse tissues. This new strategy simplifies and enhances the process of probe optimization by direct screening in a tissue of interest without any a priori knowledge of enzyme targets. It has the potential to be applied to advance the development of probes for diverse disease states for which clinical or animal model tissues are available.3