What is the oldest river on Earth?
This page is not going to tell you. Not because the answer is boring, but because the real one is better: the sources this site holds do not agree with each other, and you can watch them disagreeing inside a single organisation, in public, right now.
AI reconstruction — not a photograph · Shutterstock AI/Shutterstock.com · Shutterstock Standard License
A guess with a factor of a hundred in it
The New River runs through a gorge in West Virginia, and the national park built around it published an essay about how that gorge was cut. Here is what it says about how long it took.
“Geologists don't know, and their estimates vary widely from 3 million years to 320 million years.”
U.S. National Park Service, “Geology of New River Gorge”. Courtesy of the National Park Service.
Read that again slowly. The top of the estimate is about 107 times the bottom of it. If someone guessed your age that loosely they would say you were somewhere between one and a hundred.
That sentence is doing something unusual and worth pointing at. It is a park, writing about the river it exists to protect, on its own website, choosing to publish how uncertain the number is instead of picking the impressive end. Most places would have printed the 320.
The same agency, three different answers
Here is where it gets genuinely strange. The National Park Service does not have one position on this. It has at least three, on three of its own pages.
- 1. Hedged twice over
- 2. Shrunk to one continent
- 3. Third place, top empty
The park’s geology essay
“In fact, the New may be very old, perhaps one of the oldest river systems in the world.”
Hedged twice over — “may be”, and “perhaps one of”. Not the oldest. One of the oldest, possibly.
The same park’s front page
“among the oldest rivers on the continent”
The same river, one click away, and the claim has shrunk from the world to one continent.
A different park, about a different river
“This ancient river is believed to be the third oldest in the world, around 250 million years old.”
Not the New River, and not the same park. That page says of itself: “It is located on the east side of the Blue Ridge Parkway (Milepost 393.8) and provides views of the French Broad River.” So this is the Blue Ridge Parkway, writing about the French Broad — and it is confident, it is specific, and it hands out third place in a world ranking.
So one page will not commit to a factor of a hundred, another quietly drops from the world to the continent, and a third awards third place outright — in a race where no page found here names a first or a second. That is not a mistake anybody made. It is what an honest answer looks like when the underlying question is harder than it sounds.
Why the question is harder than it sounds
Ask how old a mountain is and you can date its rock. Ask how old a river is and the question splits apart in your hands, because a river is not a thing so much as something that keeps happening. The water is not the same water it had last week. The channel moves. The valley deepens.
The New River is a good example of the mess, because it is not even entirely a different river from an older one.
“We do know that this part of New River follows the same course followed by part of the ancient Teays River and is simply a new name for part of the Teays River system.”
U.S. National Park Service, “Geology of New River Gorge”. Courtesy of the National Park Service.
If part of a river is a renamed piece of an older river, where exactly do you start counting? That is not a puzzle this site is going to solve for you, because the source does not solve it either. It is just worth seeing that the difficulty is real rather than invented.
Both of this site’s other river pages hit the same wall and refused the same way. The Colorado has a canyon with a fairly tight date on it and a river behind that date its own source will not commit to. The Mississippi page says outright that “how old is this river” has no single answer. This page is the reason both of them stopped where they did.
Careful: the rock is not the river
This is the trap, and almost every “oldest river” claim you will read outside a park website has fallen into it. The same NPS essay describes the rock the gorge is cut through:
“more than 300 million years ago when these rock sediments were accumulating: sometimes the view is of a shore along an inland sea, sometimes the shallow bottom of the inland sea itself, and sometimes a vast swamp lush with vegetation.”
U.S. National Park Service, “Geology of New River Gorge”. Courtesy of the National Park Service.
Notice what that sentence is about: sediments piling up, hundreds of millions of years ago. It is a date for the ROCK. The other quotation, the one with the factor of a hundred in it, is a date for CUTTING THE GORGE. Those are two different events and the essay keeps them in two different paragraphs.
A knife that cuts a very old cake is not a very old knife. Somewhere between the rock date and the river date is where most of the big round numbers come from, and this page is not going to join them together to make one.
One thing that is older than every river
The same essay follows the water out of the gorge, all the way to the sea, and then says this:
“Some of this water may repeat the same trip or part of it sometime in the future, as it has in the past through the age-old cycle of evaporation, condensation, precipitation, and run-off.”
U.S. National Park Service, “Geology of New River Gorge”. Courtesy of the National Park Service.
The river might be three million years old, or it might be three hundred and twenty. The water in it has done this before, and it is very much older than either figure — old enough that the numbers get their own page.
So which river wins?
This site is not giving you a winner, and it is worth being clear that this is a decision rather than a gap. There are lists elsewhere that will name one confidently. Every source held here either hedges, shrinks its claim, or awards a place in a ranking whose top is empty.
What you can take away instead is the shape of the problem. A river’s age depends on what you decide counts as the river. The estimates for a single well-studied one span a factor of a hundred. And the organisation that looks after two of the candidates says different things about them on different pages, in public, without hiding it.
That is a much more useful thing to know than a name.
For teachers
For school
Ages: a teaching choice, not a rule. The three quotations and the diagram work from about nine; the factor-of-a-hundred sum and the rock-is-not-the-river trap suit about eleven upwards.
- The top of the estimate is about 107 times the bottom of it.
- So this is the Blue Ridge Parkway, writing about the French Broad — and it is confident, it is specific, and it hands out third place in a world ranking.
- So one page will not commit to a factor of a hundred, another quietly drops from the world to the continent, and a third awards third place outright — in a race where no page found here names a first or a second.
Classroom prompt
Why might three pages from one park service disagree? Let the class list their guesses first, then hold each guess up against the three quotations: which ones do the words themselves support, and which are only guesses?
Extension. Find an “oldest river” claim somewhere else — a book, a quiz, a website. Does its date belong to the river, or to the rock the river cuts through? The section above, “Careful: the rock is not the river”, is the test to use.
Where this comes from
- The quotations
- U.S. National Park Service: “Geology of New River Gorge” (last updated March 11, 2025), “New River Gorge National Park & Preserve • West Virginia” (last updated August 24, 2026) and “French Broad Overlook” (last updated November 1, 2024), all read . Material created by the National Park Service is in the public domain. Every quotation on this page is checked against the page it came from, word for word, each time this site is built.
- How sure is this?
- Not sure at all, and that is the subject rather than a weakness. The only arithmetic here is 320 divided by 3, on the two figures the source publishes itself. No age on this page is ours, no river is ranked by us, and where the sources disagree they are shown disagreeing rather than averaged — averaging three claims would produce a fourth that none of them made. Unlike most pages on this site, nothing here rests on a reconstruction: there is no model of the ancient world behind these numbers and no map to read them off. Every figure is an estimate a national park chose to publish, uncertainty and all, and it is reproduced with the hedges the park put on it.
Where next
How old are rivers? puts all three of this site’s river pages side by side. The other two are one click away: How old is the Colorado River?, where the canyon has a date and the river behind it does not, and How old is the Mississippi River?, which refuses a single age for the reason this page gives. And How do we know? is where this site explains why an honest answer sometimes stays uncertain — the rule this whole page follows.