The Large Hadron Collider (LHC) is the world’s largest and most powerful particle collider. Its main objectives are to explore the validity of the standard model of particle physics and to look for new physics beyond it, at unprecedented collision energies and rates. A good luminosity performance is imperative to attain these goals. In the last stage of the LHC commissioning (2011-2012), the limiting factor to achieving the design bunch spacing of 25 ns has been the electron cloud effects. The electron cloud is also expected to be the most important luminosity limitation after the first Long Shut-Down of the LHC (LS1), when the machine should be operated at higher energy and with 25-ns spacing, as well as for the planned luminosity upgrade (HL-LHC) and future high energy proton colliders (HE-LHC and VHE-LHC). This thesis contributes to the understanding of the electron cloud observations during the first run of the LHC (2010-2012), presents the first beam dynamics analysis for the next generation of high energy hadron colliders, and assists in the prediction of how electron clouds will impact the performance of the future high-luminosity and high-energy machines. In particular, the thesis discusses a method to benchmark pressure measurements at the LHC against electron cloud build-up simulations for identifying the most relevant surface parameters. This method allowed monitoring the effectiveness of LHC “scrubbing runs”, revealing that in the warm regions, the maximum Secondary Electron Yield, δmax, decreased from an initial value of about 1.9 down to about 1.2 (with a low-energy electron reflectivity R ≈ 0.3), thanks to surface conditioning. In addition, the “map formalism”, a good approximation to quickly explain and predict electron cloud effects, has been further developed and applied, for the first time, to optimize the scrubbing process at the LHC. For the HL-LHC, several novel filling schemes have been analyzed in terms of luminosity performance and electron cloud activity. Only a few of them are compatible with an electron cloud activity lower than for the baseline scenario. We highlight a promising option which could be a good fallback scenario in case the electron cloud effects prevent the injection of the baseline beam. This option could also be considered for the nominal LHC after the LS1 if electron cloud turns out to be a serious obstacle. Regarding the future high-energy proton colliders (HE- and VHE-LHC), in the frame of this thesis a performance model was developed to predict the luminosity as a function of time and to optimize the beam parameters, carrying out the first ever performance analysis for these machines. Several scenarios have been considered, including round and flat beams as well as different bunch spacings. The parameters presented in this thesis have been submitted as an input for the most recent update of the European Strategy for Particle Physics. Finally, we also report the electron cloud studies performed for both high energy machines. The large amount of primary photoelectrons generated by synchrotron radiation at these high energies motivates the consideration of high efficiency photon stops as well as other mitigation techniques (e.g. a-C coatings and clearing electrodes). Although for both machines (HE- and VHE-LHC) a tentative bunch spacing of 25 ns has been considered as baseline assumption, the results of this thesis suggest the possibility of going down to 5 ns, since such a beam would present several advantages.