In spite of considerable interest in the field, reprogramming induced pluripotent stem cells (iPSCs) directly from cancer cells has encountered considerable challenges, including the extremely low reprogramming efficiency and instability of cancer-derived iPSCs (C-iPSCs). In this study, we aimed to identify the main obstacles that limit cancer cell reprogramming. Through a detailed multidimensional kinetic optimization, a highly optimized protocol is established for reprogramming C-iPSCs using nonviral plasmid vectors. We demonstrated how the initial cancer cell density seeded could be the most critical factor ultimately affecting C-iPSCs reprogramming. We have consistently achieved an unprecedented high C-iPSC reprogramming efficiency, establishing stable colonies with typical iPSC morphology, up to 50% of which express the iPSC phenotypic (Oct3/4, Sox2, Nanog) and enzymatic (alkaline phosphatase) markers. Furthermore, established C-iPSC lines were shown to be capable of forming teratomas in vivo, containing cell types and tissues from each of the embryonic germ layers, fully consistent with their acquisition of pluripotency. This protocol was tested and confirmed in two completely unrelated human lung adenocarcinoma (A549) and mouse melanoma (B16f10) cancer cell lines and thus offers a potentially valuable method for generating effectively virus-free C-iPSCs for future applications.