Conaway and Petroni suggest that the choice of phase I design “substantially enhances the proportion of effective agents that have successful phase III trials (1).” This conclusion is based on first simulating various phase I designs aimed at determining the MTD for populations of hypothetical new agents, followed by a single-arm phase II trial of varying size, followed by a phase III trial. Unfortunately, the simulation relies on historical and outdated paradigms for development of cytotoxic antineoplastic agents, rather than well-established principles of clinical pharmacology and drug development used outside of oncology.The use of the historical oncology paradigm leads to several incorrect assumptions. The authors assume that as toxicity increases, efficacy increases. While this may be true for cytotoxic chemotherapy, it is not expected to be true for modern oncology drugs, where incremental increases in dose may have no incremental increase in efficacy and only lead to incremental toxicity due to off-target effects. In this context, the therapeutic index decreases as the dose is increased beyond that achieving maximal efficacy. The authors' failure to recognize the potential for a flat dose–response curve is particularly surprising, given that one of their cited references makes this explicit point for pembrolizumab, which they used as “a concrete example” (2).This concern is not only relevant for other immunotherapeutics, but for modern small molecules as well.A recent analysis of all patent-protected oral drugs demonstrated that the majority of drugs are labeled at too high a dose (3). Thus, despite the successful completion of the drug development process, patients may experience unnecessary toxicity due to our reliance on historical drug development paradigms that are no longer pharmacologically appropriate.In this context, I would encourage biostatisticians and others interested in the design of early clinical trials to focus their efforts on a different paradigm, one in which the optimal dose can only be determined by randomized dose-ranging phase II and III trials (4, 5). Despite the widespread use of this paradigm in other therapeutic areas, there has been little effort to modify phase I oncology designs, which should aim to define a range of doses for phase II investigation. Hopefully we can redefine success as efficacy without severe toxicity, rather than continuing to accept high rates of severe toxicity that compromise the quality of life of patients with cancer and drive up the cost of oncology care.See the Response, p. 3191M.J. Ratain is an employee of Value in Cancer Care Consortium; reports receiving commercial research grants from AbbVie and Genentech; is a consultant/advisory board member for Aptevo, Cyclacel, AbbVie, Amgen, Ascentage, BioMarin, Elion Oncology, Genentech, Portola Pharmaceuticals, and Shionogi; and reports receiving other remuneration from multiple generic pharmaceutical companies for expert testimony and patent litigation consulting. No other potential conflicts of interest were disclosed.