Wildfire is a dominant disturbance agent in pan-Arctic tundra and can significantly impact terrestrial carbon balance and ecosystem functioning. Interactions between fire and climate change can enhance their impacts on the Arctic. However, the driving mechanisms of tundra fire occurrence remain poorly understood. This study focuses on identifying key environmental factors controlling fire occurrence in Arctic tundra of Alaska. Our random forest models, considering ignition source, fuel, fire weather, and topography, have shown a strong predictive capability with an overall accuracy above 91%. We found cloud-to-ground (CG) lightning to be the dominant driver controlling tundra fire occurrence. Near-surface weather warmer and drier than normal was required to support burning, while fuel composition and topography have modest correlations with fire occurrence. Our results highlight the critical role of CG lightning in driving tundra fires and that incorporating lightning in modeling is essential for fire monitoring, forecasting, and management in the Arctic.