While over ten-thousand phase I studies are published in oncology, fewer than 1% of these studies stratify patients based on genetic variants that influence pharmacology. Pharmacogenetics-based patient stratification can improve the success of clinical trials by identifying responsive patients who have less potential to develop toxicity; however, the scientific limits imposed by phase I study designs reduce the potential for these studies to make conclusions. We compiled all phase I studies in oncology with pharmacogenetics endpoints (n = 84), evaluating toxicity (n = 42), response or PFS (n = 32), and pharmacokinetics (n = 40). Most of these studies focus on a limited number of agent classes: Topoisomerase inhibitors, antimetabolites, and anti-angiogenesis agents. Eight genotype-directed phase I studies were identified. Phase I studies consist of homogeneous populations with a variety of comorbidities, prior therapies, racial backgrounds, and other factors that confound statistical analysis of pharmacogenetics. Taken together, phase I studies analyzed herein treated small numbers of patients (median, 95% CI = 28, 24–31), evaluated few variants that are known to change phenotype, and provided little justification of pharmacogenetics hypotheses. Future studies should account for these factors during study design to optimize the success of phase I studies and to answer important scientific questions.