Fire is a natural, and necessary, component of many ecosystems. However, people are changing the spatial and temporal distribution of wildfires in the U.S. at great economic and ecological costs. My dissertation addresses the impacts of humans on U.S. fires both through the introduction of ignition sources and flammable grasses. Further, I evaluate fire datasets that are widely used to investigate these phenomena over large spatial and temporal scales. Finally, I create an aboveground carbon map that can be used to estimate the potential carbon loss consequences in western U.S. ecosystems most at risk to fire. My work shows that humans ignited more than 77% of fires in seven western U.S. ecoregions, and when modeling human ignited fires, I found that the importance of ignition proxies varied considerably among ecoregions. In 21 ecoregions across the U.S., I found that eight species of non-native invasive grasses increased rates of fire occurrence by 27%-230%, and six species increased rates of fire frequency by 24%-150%. I also quantified differences in commonly used satellite derived and agency recorded fire records and found they were disparate across the U.S., suggesting that great care should be taken when deciding which fire database to use when analyzing human impacts on fire regimes. Finally, the new estimates I provide for aboveground carbon in semi-arid western U.S. ecosystems are roughly double that of previous estimates; indicating that potential carbon losses from fire in these ecosystems are much larger than originally thought. I conclude that fire ignitions from human sources, and the alteration of fuels through the introduction of non-native, invasive grasses, have already dramatically impacted fire regimes across the U.S. These impacts are presently and will continue to be compounded by climate change. My dissertation suggests that we must consider human impacts on ignitions, vegetation, and their interaction with climate to most effectively manage, predict, and live with fire.