Identification of the optimal dose remains a key challenge in drug development. For cytotoxic drugs, the standard approach is based on identifying the maximum tolerated dose (MTD) in phase I trials and incorporating this to subsequent trials. However, this strategy does not take into account important aspects of clinical pharmacology. For targeted agents, the dose-effect relationships from preclinical studies are less obvious, and it is important to change the way these agents are developed to avoid recommending drug doses for different populations without evidence of differential antitumour effects in different diseases. The use of expanded cohorts in phase I trials to better define MTD and refine dose optimization should be further explored together with a focus on efficacy rather than toxicity-based predictions. Another key consideration in dose optimization is related to interindividual pharmacokinetic variability. High variability in intra-individual pharmacokinetics has been observed for many orally-administered drugs, especially those with low bioavailability, which might complicate identification of dose-effect relationships. End-organ dysfunction, interactions with other prescription drugs, herbal supplements, adherence, and food intake can influence pharmacokinetics. It is important these variables are identified during early clinical trials and considered in the development of further phase II and subsequent large-scale phase III studies.