As advances in radiotherapy technology enable higher precision treatments, it becomes increasingly important to understand the factors that contribute to treatment uncertainty. The recent developments in imaging modalities and computer algorithms have made possible quantitative measurements of treatment uncertainties on statistically significant numbers of patients, which has led to new strategies for reducing as well as incorporating them into the treatment planning process. This article reviews the current literature on two sources of uncertainties deemed important in photon therapy, namely, patient localization (setup) errors and organ motion. In the area of patient localization there has been increasing work on protocols using electronic portal imaging devices to correct setup errors. These protocols are derived from probability analyses based on knowledge of setup errors for a population of patients in combination with defined clinical endpoints. Measurements of organ motion and methods to correct or control it have been more limited, due partly to the larger difficulties in imaging and motion characterization. We also review two paradigms for accounting for uncertainties in treatment plans: the conventional approach, which adds a margin around the tumor volume, and an alternative one, which includes uncertainties directly in the dose distributions of the tumor volume and nearby normal organs.