Wildfires are a vital element of the hydrological and ecological systems in the western U.S. Their occurrence and severity are affected by both large-scale meteorological patterns and local hydrometeorological conditions. An understanding of the relationship between wildfire occurrence and environmental factors is important for projecting changes in wildfire occurrence in a warming climate. With a decades-long high-resolution hydroclimate simulation for 1984-2017 and observed wildfire distribution, we examine the antecedent hydrometeorological conditions including precipitation, near-surface temperature and humidity, soil moisture, and snowpack for wildfire occurrence in the western U.S. Wildfire events are classified by their controlling hydrometeorological factors, and the probability of fire occurrence associated with given hydrometeorological conditions is quantified. The predictability of wildfire based on these factors is also explored using machine learning approaches. The established prediction models are then driven by perturbed future climate simulations to estimate future wildfire risks in 2041-2070. Analysis of areas with high wildfire risks under present and future climate may be used to inform land management and risk assessment from local to regional scales.