Massive stars play an essential role in the Universe. They are rare, yet the energy and momentum they inject with their intense radiation fields and stellar winds into the interstellar medium (ISM) dwarfs the contribution by their vastly more numerous low-mass cousins. These mechanisms can halt accretion onto massive stars and limit star formation in massive star clusters (MSCs), which can host thousands of massive stars. For stellar winds, I discuss how we can use observations to constrain a range of kinetic energy loss channels for the shock-heated gas from stellar winds in MSCs. I demonstrate that the kinetic energy injected by stellar winds in MSCs is not a significant contributor to stellar feedback for young MSCs. I argue instead that radiation pressure is likely the dominant feedback mechanism in massive star and MSC formation. Therefore detailed simulation of their formation requires an accurate treatment of radiation. For this purpose, I have developed a new, highly accurate hybrid radiation algorithm that properly treats the absorption of the direct radiation field from stars and the re-emission and processing by interstellar dust. With this new method, I performed a suite of three-dimensional radiation-hydrodynamic simulations of the formation of massive stars and MSCs. For individual massive stellar systems, I find that mass is channeled to the massive stellar system via gravitational and Rayleigh-Taylor (RT) instabilities. I will also present simulations of the formation of a MSC from the collapse of a dense, turbulent, magnetized million solar mass molecular cloud. I find that the influence of the magnetic pressure and radiative feedback slows down star formation. These results suggest that the combined effect of turbulence, magnetic pressure, and radiative feedback from massive stars is responsible for the observed low star formation efficiencies in molecular clouds.