This dissertation explores one overarching question relevant to the paleoclimate of the latest Pleistocene glacial cycle (approximately the last 130,000 years): “How did spatial and temporal evolution of ocean temperature, both at the surface and interior, relate to other parts of the climate system in the late Pleistocene?” Results from three studies are presented that seek to address longstanding questions in paleoceanography and paleoclimatology for the late Pleistocene using a combination of novel and accepted statistical and geochemical analysis techniques and leveraging comparisons with available global climate model data. The last interglaciation (LIG; ~129-116 ka) was the most recent period in Earth’s history with higher-than-present global sea level (≥6-9 m) under similar-to-preindustrial concentrations of atmospheric CO₂. This suggests that additional feedbacks related to albedo, insolation, and ocean overturning circulation may have resulted in the apparent warming required to cause the higher sea level. Our understanding of how much warmer the LIG was relative to the present interglaciation remains uncertain, however, with current estimates suggesting that sea-surface temperatures (SSTs) were 0-2°C warmer than late-20th century average global temperatures. We present a global compilation of proxy-based annual SST spanning the LIG. Using Monte Carlo and Bayesian techniques to propagate uncertainties in age-model and proxy-based SST reconstructions, our results quantify the spatial timing, amplitude, and uncertainty in global and regional SST change during the LIG. Our conclusions suggest that the LIG surface ocean was indistinguishable from the average surface ocean temperatures observed for the last two decades (1995-2014). This may ultimately imply that the Earth is currently committed to ≥6-9m of equilibrium sea-level rise. Although the LIG is not an analogue for present and future climate change due to the large differences in seasonal orbital insolation and absence of anthropogenic greenhouse gas radiative forcing, it provides an opportunity to test the ability of global climate models to simulate the mechanisms and climate feedbacks responsible for the warmer climate and higher global mean sea level during the LIG. However, when forced only by LIG greenhouse gas concentrations and insolation changes, climate models suggest that the annual mean temperature response was not significantly different from preindustrial control simulations. We present the first multi-model and multiscenario ensemble of transient and equilibrium global climate modeling results spanning the LIG. We show, using a novel model-data comparison framework, that these scenario-specific model results exhibit regionally independent agreement with ocean basin-specific proxy-based SST stacks. This result ultimately implies structural uncertainties and/or misrepresentations of climate feedbacks in the existing suite of climate model simulations, or underestimations of additional proxy-based SST uncertainties. Our conclusions suggest a new target LIG time period for future model-data comparisons and highlight the need for higher resolution transient climate modeling of the LIG and its dependence on meltwater input to the high latitude oceans during the preceding deglaciation. Few discoveries have stimulated the paleoclimate community more so than Heinrich events. Nevertheless, the cause of Heinrich events, characterized by a large flux of icebergs sourced from the Hudson Strait Ice Stream into the North Atlantic, remains debated. Commonly attributed to internal ice-sheet instability, the occurrence of Heinrich events during the coldest intervals of the last glacial cycle instead suggests an external climate control. We expand on recent studies that have shown that incursions of warm subsurface waters into the intermediate depth North Atlantic Ocean destabilized an ice shelf fronting the Hudson Strait Ice Stream, causing a Heinrich event. We present new surface- and bottom-water stable isotope, trace metal, and sedimentary records from two cores taken along the Labrador margin that further support subsurface warming as a trigger of Hudson Strait Heinrich events. We further relate these changes to other sediment core records from the North Atlantic and transient deglacial climate modeling results to show that subsurface warming was ubiquitous across the intermediate North Atlantic during the early part of the last deglaciation and was most likely caused by a preceding reduction in the Atlantic Meridional Overturning Circulation.