Tight ligand-receptor binding, paradoxically, is a major root of drug resistance in cancer chemotherapy. To address this problem, instead of using conventional inhibitors or ligands, we focus on the development of a novel process—enzyme-instructed self-assembly (EISA)—to kill cancer cells selectively. Here we demonstrate that EISA as an intracellular process to generate nanofibrils of short peptides for selectively inhibiting cancer cell proliferation, including drug resistant ones. As the process that turns the non-self-assembling precursors into the self-assembling peptides upon the catalysis of carboxylesterases (CES), EISA occurs intracellularly to selectively inhibit a range of cancer cells that exhibit relatively high CES activities. More importantly, EISA inhibits drug resistant cancer cells (e.g., triple negative breast cancer (TNBC) cells (HCC1937) and platinum-resistant ovarian cells (SKOV3, A2780cis)). With the IC50 values of 28–80 μg/mL and 25–44 μg/mL of L- and D-dipeptide precursors against cancer cells, respectively, EISA is innocuous to normal cells. Moreover, using co-culture of cancer and normal cells, we validate the selectivity of EISA against cancer cells. Besides revealing that intracellular EISA cause apoptosis or necroptosis to kill the cancer cells, this work illustrates a new approach to amplify the enzymatic difference between cancer and normal cells and to expand the pool of drug candidates for potentially overcoming drug resistance in cancer therapy.