People start the vast majority of forest fires. Across the contiguous United States, human activity accounts for roughly 84% of all wildfire ignitions, while lightning accounts for the rest. But that split tells only part of the story, because lightning-caused fires tend to burn in remote, rugged terrain where they can grow large, and they still account for about 44% of the total area burned. The balance between natural and human causes shifts dramatically depending on where you are, what time of year it is, and what the landscape looks like.
Lightning as the Primary Natural Cause
Lightning is, for practical purposes, the only significant natural ignition source for forest fires. Volcanic eruptions and spontaneous combustion of organic material can technically start fires, but they are vanishingly rare compared to the billions of cloud-to-ground lightning strikes that hit the planet each year. The strikes that matter most for wildfire are the ones accompanied by little or no rain, commonly called “dry lightning.” A bolt hits a tree or dry ground cover, the heat ignites the fuel, and there is not enough moisture falling to snuff the fire out.
The conventional rule of thumb has been that lightning striking with less than 2.5 millimeters of daily rainfall qualifies as dry lightning and poses ignition risk. Research combining wildfire, lightning, and precipitation datasets across the western United States found the reality is more complicated. The median rainfall accompanying lightning-ignited wildfires was actually 2.8 mm, and it varied considerably by vegetation type and region. Fires detected promptly after a strike occurred with a median of about 2.5 mm of rain, but so-called “holdover” fires, which smolder undetected for two to five days before flaring up, occurred with significantly more precipitation, around 5.1 mm. Across different ecological regions, the range for promptly detected fires ran from 1.7 to 4.6 mm, meaning the single 2.5 mm cutoff misses a lot of real-world ignition risk.1Geophysical Research Letters. Lightning‐Ignited Wildfires in the Western United States: Ignition Precipitation and Associated Environmental Conditions
Those holdover fires deserve extra attention. A lightning bolt strikes during a storm, the initial flame gets dampened by rain, but the fire creeps into duff, roots, or deep organic soil and smolders invisibly for days. When conditions dry out again, the fire re-emerges. This makes holdover fires especially difficult to detect and manage, because by the time anyone spots smoke, the origin strike may have happened nearly a week earlier under completely different weather. Smouldering fires in peat and deep organic soils can persist for remarkably long periods and are responsible for some of the largest fires on Earth, particularly in boreal and tropical peatlands.2PubMed Central. Smouldering wildfires in peatlands, forests and the arctic: Challenges and perspectives
How Human Activity Dominates the Numbers
A landmark analysis of U.S. wildfire records found that humans account for 84% of all ignitions and 44% of total area burned. Human-caused fires were the dominant source in over 5.1 million square kilometers of the country, while lightning dominated in only about 0.7 million square kilometers, mostly in the sparsely populated mountainous West.3PubMed Central. Human-started wildfires expand the fire niche across the United States The gap between 84% of ignitions and 44% of area burned reflects an important pattern: human-caused fires tend to start closer to roads, towns, and agricultural land, where they are usually spotted and suppressed faster. Lightning fires, by contrast, often start in wilderness areas where detection is slower and access for firefighters is harder.
Human ignition sources are diverse. They include campfires left unattended, debris burning that escapes, arson, fireworks, sparks from vehicles and equipment, agricultural burning, and electrical infrastructure failures. A modeling study that tried to assign causes to over 150,000 U.S. wildfires between 1992 and 2020 whose ignition source was officially listed as unknown attributed the largest share to equipment and vehicle use (21%), followed by lightning (20%) and arson (18%).4Earth’s Future. Inference of Wildfire Causes From Their Physical, Biological, Social and Management Attributes That equipment category covers everything from chainsaws and mowers throwing sparks to hot exhaust pipes parked over dry grass.
Another human cause that generates a predictable annual spike is fireworks. U.S. wildfire data show a surge of ignitions every Fourth of July, with fireworks-caused fires concentrated in the western and north-central states and disproportionately on tribal lands.5PubMed Central. The timing of fireworks-caused wildfire ignitions during the 4th of July holiday season It is one of the most geographically and temporally predictable wildfire patterns in the country.
Agricultural fire use is a major cause globally. In regions like the Peruvian Andes, rural communities have used fire for land preparation, weed control, and soil management since long before modern regulation. Despite strict legal prohibitions, wildfire frequency in those areas has actually increased, suggesting that punitive policies alone do not solve the problem when fire is deeply embedded in agricultural practice.6ScienceDirect / Elsevier (International Journal of Disaster Risk Reduction). Rethinking the agricultural use of fire and its influence on the occurrence of wildfire in high Andean communities of Cusco, Peru
Regional Patterns Matter More Than National Averages
National statistics on fire causes can be misleading if you do not break them down by region. In the western United States, lightning and campfires are the dominant ignition sources, reflecting the combination of dry summer thunderstorms and heavy recreational use of public lands. East of the Mississippi River, the picture flips: arson is the leading cause of wildfire. Housing growth explains more of the variation in fire frequency, while climate conditions primarily influence how much area burns once a fire starts.7Environmental Research Letters. Regional patterns in U.S. wildfire activity: the critical role of ignition sources
The same geographic logic applies outside the United States. In southeastern Australia, researchers modeling the drivers of wildfire ignitions found that fire weather conditions had a consistently positive effect on ignition likelihood, but weather contributed far more to lightning-caused fires (57%) and power-transmission fires (55%) than to other human-caused categories, which ranged from 8% to 32%.8PubMed. Developing and testing models of the drivers of anthropogenic and lightning-caused wildfire ignitions in south-eastern Australia In other words, weather amplifies all fire causes, but some human ignitions happen regardless of conditions, while lightning fires and power-line fires are almost entirely weather-dependent.
Power Lines and Electrical Infrastructure
Electrical infrastructure is responsible for some of the most destructive wildfires in recent memory. The mechanism is straightforward: power lines sag in heat, swing in wind, or fail when trees fall on them. Contact between a live conductor and vegetation or dry surface fuel can produce an ignition instantly. The heating process on the conductor surface itself can also transfer enough energy to start a fire even without direct contact with a tree.9International Journal of Electrical Power & Energy Systems. Preignition risk mitigation model for analysis of wildfires caused by electrical power conductors
Utilities in fire-prone areas have increasingly adopted public safety power shutoffs, preemptively cutting electricity to high-risk zones during extreme fire weather, as a blunt but effective tool to prevent ignitions. The tradeoff is obvious: shutting off power to tens of thousands of homes and businesses causes significant economic and quality-of-life disruption. This has driven a push for underground power lines, covered conductors, and better real-time monitoring of line conditions. The cost of these upgrades runs into the billions, but so do the damages from a single catastrophic fire.
Fuel Moisture, Wind, and Why a Spark Is Not Enough
An ignition source alone does not make a forest fire. The fuel has to be dry enough to catch, and conditions have to support the fire’s spread. Fuel moisture content is one of the most important variables in determining whether a spark turns into a wildfire or fizzles out. Research in northeast China’s forests found that some fuel types would not ignite even at 15% moisture content, while others could catch fire and sustain spread at moisture levels as high as 38% to 40%.10Journal of Forestry Research. Moisture content thresholds for ignition and rate of fire spread for various dead fuels in northeast forest ecosystems of China The variability depends on the physical structure of the fuel: fine, loosely packed material like dry grass ignites easily, while dense woody debris requires more energy and drier conditions.
At the landscape level, different vegetation types have distinct moisture thresholds that signal fire danger. Research aimed at building operational early warning systems established that extreme fire hazard in grasslands occurs when fuel moisture drops below about 55%, while forests reach extreme hazard around 72% and shrublands around 106%.11PubMed Central. Determining fuel moisture thresholds to assess wildfire hazard: A contribution to an operational early warning system Shrublands tolerate higher moisture because their leaves contain volatile oils that burn readily even when the plant is not completely dried out.
Wind is the other critical accelerant. Southern California’s Santa Ana wind events provide a vivid case study. These hot, dry offshore winds create extreme fire danger every autumn. Analysis of fires during Santa Ana events showed that while moderate winds were responsible for the majority of fires and area burned, extreme wind days could produce enormous individual fires. The 2017 Thomas Fire, which was at the time the largest in modern California history, coincided with a Santa Ana event that lasted 16 days and included four extreme wind days, a combination that was without precedent in 71 years of records. Yet other events with multiple extreme wind days produced no fire activity at all, which illustrates a key point: wind creates the potential, but an ignition source still has to be present.12PubMed Central. Ignitions explain more than temperature or precipitation in driving Santa Ana wind fires The researchers found that the number and location of ignitions explained fire outcomes during wind events better than temperature or precipitation did.
Climate Change Is Adding Lightning Days
A warming atmosphere holds more moisture and energy, which means more convective storms and, in turn, more lightning. Climate projections using Earth system models find widespread increases in the number of cloud-to-ground lightning days across the western United States by the mid-21st century. The increases are especially pronounced in the northern West, where many areas could experience 4 to 12 additional lightning days per year compared to the period from 1995 to 2022.13Earth’s Future. Projections of Lightning‐Ignited Wildfire Risk in the Western United States Climate simulations also project a roughly 18% increase in lightning activity driven by atmospheric destabilization and increased ice particle formation in storm clouds.14PubMed Central. Significant increase in graupel and lightning occurrence in a warmer climate simulated by prognostic graupel parameterization
More lightning days in a landscape that is simultaneously getting hotter and drier is a compounding problem. The same warming that generates extra thunderstorms also dries out fuels faster and extends the window during which vegetation is flammable. This is why projections for lightning-ignited wildfire risk show increases beyond what lightning frequency alone would predict. Regions that historically had short fire seasons, such as parts of the northern Rockies and the Pacific Northwest, are expected to see the most dramatic shifts.
A Century of Fire Suppression and Its Consequences
For most of the 20th century, the dominant fire management policy across North America was total suppression: put out every fire as fast as possible. This approach, while effective at protecting property in the short term, fundamentally altered the landscapes it was meant to protect. Decades of suppression allowed fuels to accumulate in forests that historically burned regularly, leading in some areas to unnaturally severe wildfires when fires eventually did escape control. In other areas, fire exclusion shifted the balance toward more fire-sensitive tree species, changing the character of the forest itself.15Frontiers in Ecology and the Environment. Prescribed fire in North American forests and woodlands: history, current practice, and challenges
This history matters for understanding modern fire causes because it changed what happens after ignition. A lightning strike in a forest that burned naturally every 10 to 15 years would have encountered light fuels and produced a low-intensity surface fire. The same strike in a forest that has not burned in 80 years encounters a dense understory, heavy duff, and ladder fuels that carry fire into the canopy. The cause of the fire is the same; the outcome is vastly different.
Before European colonization, Indigenous peoples across North America used fire extensively as a land management tool. Research in the Karuk Aboriginal Territory in northern California estimated roughly 6,972 annual cultural burning ignitions in that landscape alone, averaging about 6.5 ignitions per fire steward per year.16PubMed. Blending Indigenous and western science: Quantifying cultural burning impacts in Karuk Aboriginal Territory These fires shaped the composition, structure, and resilience of ecosystems over thousands of years. The suppression era eliminated both natural and cultural fire from many landscapes simultaneously, and restoring fire to its historical role is now a central challenge in wildfire management.
Plants That Evolved to Burn
Fire is not just a destructive force that forests endure. Many plant species have evolved traits that depend on fire, and their presence on a landscape can influence how fires behave once they start. This relationship between fire and vegetation goes back hundreds of millions of years.
Serotiny, the trait in which seeds are stored in sealed cones on the plant and only released after heat from a fire melts the resin seal, is one of the best-known fire adaptations among conifers. Researchers tracing this trait’s evolutionary history have proposed that serotiny has been expressed in conifers for roughly 350 million years, making it one of the oldest fire-linked adaptations known. The trait requires specific structural features: a woody support, compact protective scales, a seed wing for dispersal after release, and crown fire as the selective pressure that triggers seed release.17Journal of Ecology. A 350‐million‐year legacy of fire adaptation among conifers
Other species take a different approach. Obligate seeders, like certain chaparral shrubs in California, do not resprout after fire. Instead, they recruit profusely from seed banks that germinate only in post-fire conditions. Early botanical descriptions from the 1920s noted that some of these species “adapt themselves to short fire intervals” by fruiting within five or six years and reappearing promptly after burns.18PubMed Central. Evolutionary fire ecology: An historical account and future directions Some Mediterranean pines go further and actively promote fire through their own chemistry. Their needles contain volatile terpene compounds that increase flammability, and the concentration of these compounds varies with fire frequency in the local environment, suggesting a feedback loop between fire and the chemical traits of the trees that live with it.19PubMed Central. Effect of Fire Frequency on the Flammability of Two Mediterranean Pines: Link with Needle Terpene Content
This evolutionary dimension complicates the simple framing of “what starts forest fires.” In fire-adapted ecosystems, the vegetation itself is part of the ignition equation. A landscape filled with resinous, volatile-rich species will burn more readily from a smaller ignition source than a landscape of less flammable hardwoods. The fuel and the fire have been shaping each other for longer than humans have existed.
Why So Many Fires Are Listed as “Cause Unknown”
A surprisingly large fraction of wildfires end up with no officially determined cause. In the U.S. federal fire reporting system, the “unknown” or “miscellaneous” category has historically been one of the largest. Determining what started a fire requires investigators to locate the exact point of origin in terrain that may have burned over completely, then find physical evidence, such as a match, a piece of glass, melted power-line hardware, or a lightning-scarred tree, that ties a specific ignition mechanism to that point. In remote areas, the point of origin may never be visited at all.
Researchers have tried to fill this gap using statistical modeling. By examining the physical setting, weather conditions, nearby land use, time of year, and management history of fires with known causes, they built models to infer the likely causes of fires labeled unknown. As noted earlier, this approach attributed the largest shares to equipment and vehicles, lightning, and arson.4Earth’s Future. Inference of Wildfire Causes From Their Physical, Biological, Social and Management Attributes Satellite-based detection is also improving the picture for lightning-caused fires. In central Brazil, researchers combined ground-based lightning detection networks with satellite fire-mapping sensors to identify fires that occurred close in time and space to cloud-to-ground strikes, building a more systematic record of natural ignitions in a region where ground-based investigation is often impractical.20Agricultural and Forest Meteorology. Characteristics of lightning-caused wildfires in central Brazil in relation to cloud-ground and dry lightning
Getting the cause right matters beyond satisfying curiosity. Prevention strategies differ completely depending on what is starting fires. If a region’s fires are mostly caused by arson, the response is law enforcement and community engagement. If they come from power lines, the answer is infrastructure upgrades or shutoffs. If lightning is the dominant cause, the most productive investment may be in early detection and rapid response rather than prevention, since you cannot stop lightning. And if escaped agricultural burns are the problem, the evidence from places like Peru suggests that criminalization alone does not work and that integrating controlled fire into land management may be more effective than banning it.