In this study, we highlighted the importance of molecular 3D similarity calculations in drug design by discovering a novel pteridin-7(8H)-one scaffold for irreversible EGFR inhibitors using a computational scaffold-hopping strategy. To discover irreversible inhibitors with a novel scaffold based on previous research on the conversion of a pyrimidine scaffolds-based EGFR inhibitor to an irreversible EGFR T790M mutant inhibitor, an in-house molecular 3D similarity calculation program, SHAFTS, was used to search the NCI database for an alternative scaffold. This search led to the discovery of a novel pteridin-7(8H)-one EGFR-inhibitor scaffold. As expected, further conversion of the new scaffold into an irreversible inhibitor was achieved by introducing Michael addition receptor groups. This addition led to highly potent EGFR T790M mutant inhibitors that exhibited competitive inhibitory enzymatic activities with subnanomolar IC50 values against both wild-type and T790M/L858R mutant EGFRs. The most active compounds, 3q and 3x, strongly inhibited the growth of gefitinib-resistant H1975 cells. Kinase-selectivity profiling revealed that one of the most effective compounds, 3x, is highly selective to EGFR and corresponding mutants, indicating a good safety index. Moreover, an in vivo antitumor efficacy study demonstrated that compound 3x significantly inhibited tumor growth and induced tumor stasis in an EGFR-T790M/L858R-driven human NSCLC xenograft mouse model of H1975 by IP dosing at 20 mg/kg/day. In conjunction with the binding mode prediction data, these results suggest that the novel pteridin-7(8H)-one-based inhibitors utilize the same mode of action as the patent- protected pyridine inhibitors. Our investigations demonstrate the pivotal roles of molecular shape and chemotype and matching in scaffold hopping and demonstrate the use of molecular 3D calculations to provide a novel starting point with which to generate competitive bioactivities.