Cancer cells produce elevated levels of reactive oxygen species, which has been used to design cancer specific prodrugs. Their activation relies on at least a bimolecular process, in which a prodrug reacts with ROS. However, at low micromolar concentrations of the prodrugs and ROS, the activation is usually inefficient. Herein, we propose and validate a potentially general approach for solving this intrinsic problem of ROS‐dependent prodrugs. In particular, known prodrug 4‐(N‐ferrocenyl‐N‐benzylaminocarbonyloxymethyl)phenylboronic acid pinacol ester was converted into its lysosome‐specific analogue. Since lysosomes contain a higher concentration of active ROS than the cytoplasm, activation of the prodrug was facilitated with respect to the parent compound. Moreover, it was found to exhibit high anticancer activity in a variety of cancer cell lines (IC50=3.5–7.2 μm) and in vivo (40 mg kg−1, NK/Ly murine model) but remained weakly toxic towards non‐malignant cells (IC50=15–30 μm).