Tropical Forest Loss Fell 36%. It Is Still 46% Higher Than a Decade Ago.

New data from the University of Maryland's GLAD Lab, published on the World Resources Institute's Global Forest Watch platform, reports that tropical primary rainforest loss fell 36% in 2025.

The world lost 4.3 million hectares of tropical primary rainforest last year — about the area of Denmark.

That is a large fall, and it is real. Here are the two sentences that must sit next to it:

2024 was the worst year ever recorded. A 36% fall from a record high is a fall back toward the range, not a fall out of it.

Loss is still 46% higher than a decade ago. Primary forest is still disappearing at roughly eleven football pitches a minute.

The shape of the good news

The recovery is concentrated in one factor: fire.

Fire-driven tropical primary forest loss fell 50% in 2025 against 2024, which had seen record-breaking fires across the tropics, especially in Latin America.

So most of the 36% headline is the tropics simply not burning the way they burned in 2024 — and 2024's burning was itself driven by drought conditions that eased. That is weather doing the work, not policy. It can reverse next year without anyone changing a single law.

WRI's own framing is blunt: climate-driven fires are a dangerous new normal that threatens to reverse recent gains.

The good news that isn't weather

There is a second number underneath, and it is the durable one.

Brazil cut non-fire primary forest loss by 41% in 2025 — its lowest level on record.

Read the qualifier: non-fire. That is clearing. Chainsaws, bulldozers, conversion. It is the part a government can actually control, and Brazil controlled it, coinciding with the relaunch of the PPCDAm federal anti-deforestation plan and stiffer penalties for environmental crime.

Weather gave the world the 36%. Policy gave Brazil the 41%. Only one of those is repeatable on purpose.

This corroborates what we reported from Brazil's own INPE satellites in Amazon Deforestation Is Falling — two independent systems, one Brazilian and one American, agreeing on direction. That agreement matters more than either number alone.

Why this number isn't the same as Brazil's number

If you compare this article's figures with INPE's, they will not match, and it is worth knowing why before someone accuses somebody of cooking books.

• INPE PRODES measures clearing in the Brazilian Amazon, on an August–July year. It tracks deforestation separately from fire loss.
• GLAD / Global Forest Watch measures primary forest loss across the whole tropics, on a calendar year, and includes fire.

Different geography, different calendar, different definition of loss. Both are correct. Neither is the other's check.

This is the same lesson as the 9.6%-versus-3.2% ocean protection gap: when two credible sources disagree, look at what each one is counting before deciding who is wrong. Usually nobody is.

Bolivia

The single most striking figure in the release has nothing to do with the headline.

Bolivia recorded its second-highest primary forest loss on record and now ranks second in the world for tropical primary forest loss — ahead of the Democratic Republic of the Congo.

Bolivia has about 60% less primary forest than the DRC.

Sit with the arithmetic. A country with a fraction of the forest is losing more of it, in absolute terms, than the nation that holds the second-largest rainforest on Earth. Driven by severe fires in 2024 whose effects carried.

A country can only do that for so long. Bolivia is not a large forest nation losing a small share; it is a smaller forest nation losing a large one. That is a different and faster kind of trajectory, and it is happening while global attention is on the Amazon's good news next door.

What "primary forest" means, and why it's the right unit

Primary forest is mature, largely undisturbed natural forest — not plantation, not regrowth, not tree cover in the generic satellite sense.

The distinction is doing real work. Tree cover loss counts any tree, including a harvested plantation that will be replanted next season. Primary forest loss counts the irreplaceable kind: the forest with the deep carbon, the intact species assemblages, the thing that does not come back on a human timescale.

A hectare of primary forest replaced by a hectare of plantation is, to a sufficiently naive satellite metric, no loss at all. It is a catastrophic one. This dataset is designed to see that difference, which is why we cite it rather than headline tree-cover figures.

Why this is on AEDO1

The honest summary of 2025 is uncomfortable to write and easy to distort in either direction.

Forest loss dropped sharply, mostly because it rained. Brazil earned a real win on the part it controls. Bolivia is in serious trouble and almost nobody is covering it. And the total — 4.3 million hectares of irreplaceable forest, in the good year — is still half again what it was ten years ago.

There is no version of that paragraph that makes a satisfying headline. It is what the data says.