Background: Patients recruited in phase I oncology trials are often treated at doses lower than the maximum tolerated dose (MTD), and therefore may not receive the most efficacious dose available, despite their expectations to the contrary. This report investigates the consequences of allowing a patient choice of dose within a common dose-escalation scheme. Methods: Trials using the continual reassessment method of dose escalation are simulated, with a modification of the rules to allow patients to choose a higher dose if they wish. The effect of allowing this choice is assessed in terms of probability of toxicity and probability of being treated at the MTD or higher. Results: The simulations show that allowing a patient choice of dose reduces the proportions of patients treated at doses lower than the MTD, and has little impact on the overall probability of correct identification of the MTD. Conclusion: The results illustrate the principle that a choice of dose can be offered to patients in such trials without compromising the overall properties of the study.