High-performance operation in tokamaks is characterized by the formation of a pedestal, a region of suppressed transport and steep gradients in density, temperature, and pressure near the plasma edge. The pedestal height is strongly correlated with overall fusion performance, as a substantial pedestal supports the elevated core pressure necessary for the desired fusion reaction rate and power density. However, stationary operation requires some relaxation of the particle transport barrier, to avoid the accumulation of impurities (e. g., helium "fusion ash," plasmafacing surface materials) in the plasma. Moreover, the formation of the pedestal introduces an additional constraint: the steep gradients act as a source of free energy for Edge-Localized Mode (ELM) instabilities, which on ITER- or reactor-scale devices can drive large, explosive bursts of particle and energy transport leading to unacceptable levels of heat loading and erosion damage to plasma-facing materials. As such, the suppression, mitigation, or avoidance of large ELMs is the subject of much current research. In light of this, a firm physical understanding of the pedestal structure and stability against the ELM trigger is essential for the extrapolation of high-performance regimes to large-scale operation, particularly in operating scenarios lacking large, deleterious ELMs. This thesis focuses on the I-mode, a novel high-performance regime pioneered on the Alcator C-Mod tokamak. I-mode is unique among high-performance regimes in that it appears to decouple energy and particle transport, reaching H-mode levels of energy confinement with the accompanying temperature pedestal while maintaining a L-mode-like density profile and particle transport. I-mode exhibits three attractive properties for a reactor regime: (1) I-mode appears to be inherently free of large ELMs, avoiding the need for externally-applied ELM control. (2) The lack of a particle transport barrier maintains the desired level of impurity flushing from the plasma, avoiding excessive radiative losses. (3) Energy confinement in I-mode presents minimal degradation with input heating power, contrary to that found in H-mode. This thesis presents the results from a combined empirical and computational study of the pedestal on C-Mod. Analysis methods are first implemented in ELMy H-mode base cases on CMod -- in particular, the EPED model based on the combined constraints from peeling-ballooning MHD instability and kinetic-ballooning turbulence is tested on C-Mod. Empirical results in ELMy H-mode are consistent with the physics assumptions used in EPED, with the pedestal pressure gradient constrained by [delta]p ~ I2/p expected from the ballooning stability limit. To lowestorder approximation, ELMy H-mode pedestals are limited in [beta]p,ped, with the attainable beta set by shaping -- within this limit, an inverse relationship between pedestal density and temperature is seen. The pedestal width is found to be described by the scaling [delta][psi] = G[beta] 1/2 / p.ped expected from the KBM limit, where G([nu],[epsilon], ...) is a weakly varying function with hGi = 0.0857. No systematic secondary scalings with field, gyroradius, shaping, or collisionality are observed. The EPED model, based on these assumptions, correctly predicts the pressure pedestal width and height to within a systematic ~20% uncertainty. Empirical scalings in I-mode highlight the operational differences from conventional H-modes. The temperature and pressure pedestal exhibit a positive trend with current, similar to H-mode (although I-mode pedestal temperature typically exceeds that found in comparable H-modes) -- however, the temperature and pressure respond significantly more strongly to heating power, with Te ... The I-mode density profile is set largely independently of the temperature pedestal (unlike ELMy H-mode), controlled by operator fueling. Given sufficient heating power to maintain a consistent ..., temperature pedestals are matched across a range of fueling levels. This indicates a path to readier access and increased performance in Imode, with the mode accessed at moderate density and power, after which the pedestal pressure is elevated with matched increases in fueling and heating power. Global performance metrics in I-mode are competitive with H-mode results on C-Mod, and are consistent with the weak degradation of energy confinement with heating power. I-mode pedestals are also examined against the physics basis for the EPED model. Peelingballooning MHD stability is calculated using the ELITE code, finding the I-mode pedestal to be strongly stable to the MHD modes associated with the ELM trigger. Similarly, modeling of the KBM using the infinite-n ballooning mode calculated in BALOO as a surrogate for the threshold indicates that the I-mode pedestal is stable to kinetic-ballooning turbulence, consistent with the observed lack of a trend in the pedestal width with [beta]p,ped. This is found to be the case even in I-modes exhibiting small, transient ELM-like events. The majority of these events are triggered by the sawtooth heat pulse reaching the edge, and do not negatively perturb the temperature pedestal -- it is proposed, then, that these events are not true peeling-ballooning-driven ELMs, but rather are an ionization front in the SOL driven by the sawtooth heat pulse. There are transient ELM events showing the characteristic temperature pedestal crash indicating a true ELM -- the steady I-mode pedestals around these isolated events are also modeled to be P-B and KBM stable, although more detailed modeling of these events is ongoing.