What Is the Oldest Civilization in the World?

The standard answer is Sumer, in southern Mesopotamia — present-day Iraq — where cities, writing and state administration appear together earlier than anywhere else on record. Settlements there run back to around 5400 BC, and by roughly 3400 BC the city of Uruk was producing written accounts on clay. The caveat is that "oldest civilisation" is not really an archaeological question; it is a definitional one, and the answer changes completely depending on which criterion you decide matters.

That sounds like a dodge, but it is the most interesting thing about the topic. Push on the word "civilisation" and a stranger picture appears: monumental stone architecture raised four thousand years before the first city, pyramid-scale construction in Peru by people who had no pottery, and a vast urban society in the Indus Valley whose writing nobody can read.

What counts as a civilisation?

Most definitions in use today descend from a 1950 paper by the Australian archaeologist V. Gordon Childe, "The Urban Revolution", which set out ten traits separating cities from villages. The list has been trimmed and rephrased since, but the usual criteria are:

• Cities — dense, permanent settlements substantially larger than a village
• Writing or a comparable record-keeping system
• Specialised labour — people who are not farmers: potters, scribes, soldiers, priests
• Monumental architecture — building well beyond household need
• State organisation — centralised authority, taxation, law, administration
• Social stratification — durable differences in wealth and status
• Long-distance trade and the agricultural surplus to fund all of the above

Apply all seven strictly and Sumer wins. Relax any one of them and the field opens up, which is why no two lists of "oldest civilisations" agree.

There are also researchers who think the criteria are the problem. The word "civilisation" carries a value judgement inherited from the nineteenth century, when it was used to rank societies on a ladder with industrial Europe at the top. David Graeber and David Wengrow argued in The Dawn of Everything (2021) that the assumed sequence — foraging, then farming, then villages, then cities, then states — does not match the archaeological record. Their reading drew heavy criticism, but the underlying observation is widely shared: the progression taught in schools is tidier than the evidence.

Keep that in mind below. The dates are approximate, they shift as new sites are excavated, and the rankings depend on the rule you picked before you started counting.

Sumer: the usual answer

Southern Mesopotamia, the floodplain between the Tigris and Euphrates, is where the full package first appears in one place.

Farming villages show up there from around 5500 BC in what archaeologists call the Ubaid period. Eridu, remembered by the Sumerians themselves as the first city, has occupation layers from roughly 5400 BC. Over the following two millennia, settlements consolidated into something genuinely urban: by around 3100 BC, Uruk may have held somewhere between 25,000 and 50,000 people, which would make it the largest settlement on Earth at the time. Figures like these are reconstructions from surveyed house plots, so treat them as a range rather than a census.

The decisive evidence is the writing. Clay tablets from the Uruk IV levels, dated to roughly 3400–3200 BC, carry proto-cuneiform: pictographic signs pressed into damp clay, later simplified into the wedge-shaped script that gives cuneiform its name. The content is the striking part. These are not poems, prayers or royal boasts but receipts — quantities of barley, jars of beer, headcounts of sheep, rations issued to workers, the officials responsible. The earliest writing in the world is accounting, invented because a temple bureaucracy had more transactions than anyone could remember.

That detail tells you what kind of society this was: one with enough surplus to redistribute and enough hierarchy to need a record of who owed what. Monumental temple platforms rose at Uruk and Eridu through the fourth millennium BC — the Anu Ziggurat and its White Temple date to around 3500–3000 BC. The tall stepped ziggurats most people picture, like the one at Ur, are considerably later, built around 2100 BC.

Cities, writing, priests and scribes, monuments, centralised administration: on the conventional criteria, Sumer is the earliest place where they converge.

Ancient Egypt

Egypt is the perennial co-claimant, and the tie is closer than most summaries admit.

The Nile valley had substantial farming communities through the fourth millennium BC — the Naqada cultures, roughly 4000 to 3100 BC, with craft specialisation, elite burials and trade reaching into the Levant and Nubia. Unification of Upper and Lower Egypt is conventionally dated to around 3100 BC and associated with a king named Narmer, though whether Narmer and the semi-legendary Menes were one person, and whether unification was an event or a drawn-out process, remain open questions.

Egyptian writing is nearly as old as Mesopotamian. Bone and ivory labels from Tomb U-j at Abydos, dated to roughly 3250–3200 BC, carry hieroglyphic signs recording quantities and place names — administrative tags, much like the Uruk tablets. Whether hieroglyphs were invented independently or prompted by contact with Mesopotamia is genuinely unsettled. The two scripts share no signs and work on different principles, which argues for independence; the near-simultaneous appearance and documented trade contact keep the stimulus-diffusion argument alive.

So Egypt often ties with Sumer: a century or two behind on writing, slightly ahead on large-scale political unification. Where Egypt clearly wins is continuity and legibility — a pharaonic tradition running roughly three thousand years, with a decipherable language and an enormous documentary record. Sumer's cities were absorbed by Akkadians, Babylonians and Assyrians; the Egyptian state kept its own name for its entire run.

The Indus Valley

The Indus Valley civilisation — often called Harappan, after the first site excavated — covered a larger area than Sumer and Egypt combined, across what is now Pakistan and north-western India.

Its roots go deep: Mehrgarh, in Balochistan, shows farming and herding from around 7000 BC. Early Harappan settlements appear from roughly 3300 BC, and the Mature Harappan phase runs from about 2600 to 1900 BC, when Harappa and Mohenjo-daro each may have housed tens of thousands of people.

The urban planning is the headline. Streets laid on grids, fired bricks manufactured to a standard 1:2:4 ratio across hundreds of kilometres, covered drains beneath the streets, wells, public bathing structures including the Great Bath at Mohenjo-daro, and standardised weights implying regulated trade. The engineering is arguably more systematic than anything contemporary in Mesopotamia or Egypt.

What is missing is equally striking. No structure has been convincingly identified as a palace, and there are no royal tombs, no monumental depictions of rulers, no clear evidence of a standing army. Some archaeologists read this as a society organised without the kind of kingship found elsewhere; others argue the absence of a recognised palace is not proof there was none.

Then there is the script. Thousands of short Indus inscriptions survive, mostly on stamp seals, averaging around five signs, with a sign inventory usually estimated in the high hundreds. Nobody has deciphered it, and there is no bilingual text like the Rosetta Stone to work from. A minority of scholars — notably Steve Farmer, Richard Sproat and Michael Witzel in 2004 — have argued the symbols are not writing at all but a non-linguistic sign system. That claim is disputed, but it makes the point: if writing is your criterion, the Indus poses a problem nobody is resolving soon.

China

China is one of the independent cradles, though its earliest layers are messier than the traditional dynastic story suggests.

Neolithic farming cultures along the Yellow and Yangtze rivers go back to around 7000–5000 BC, with Yangshao (roughly 5000–3000 BC) and Longshan (roughly 3000–1900 BC) showing craft specialisation, walled towns and elite burials. The Liangzhu culture in the Yangtze delta, roughly 3300–2300 BC, built a walled city with an enormous system of dams and water-management works, and produced finely worked jade in quantities implying both specialists and stratification. UNESCO inscribed Liangzhu in 2019 explicitly as evidence of an early state — an argument that would have been unusual a few decades ago.

Erlitou, in Henan, dates to roughly 1900–1500 BC and has palace foundations, bronze casting and clear urban planning. It is frequently identified with the Xia, the first dynasty named in later Chinese texts, but that identification is contested: the archaeology is solid, the link to a dynasty recorded a thousand years afterwards is an interpretation.

Secure written Chinese arrives with the Shang, conventionally about 1600–1046 BC. The oracle bones from Anyang, from roughly 1250 BC onward, are divination records inscribed on ox scapulae and turtle shells — questions put to ancestors about harvests, warfare, childbirth and the weather. The script is directly ancestral to modern Chinese characters, which is why China claims the longest continuously used writing system, even though it appears some two thousand years after Sumer's.

The Americas

Complex society in the Americas developed with no contact with the Old World, which makes it the best natural experiment available for testing what is actually necessary to build one.

The Norte Chico or Caral-Supe civilisation, on the arid Peruvian coast north of Lima, runs from roughly 3500 to 1800 BC, with monumental construction at Caral from around 2600 BC — putting its large platform mounds and sunken circular plazas broadly contemporary with the Egyptian pyramids at Giza. Around thirty related sites have been identified in the river valleys of the region.

What Norte Chico did not have is the interesting part. No pottery has been found, and no writing, though a knotted cord recovered at Caral has been suggested as a precursor to the Andean quipu — a proposal, not a settled finding. There is little evidence of warfare: no fortifications, no weapons caches, no mass graves. The economy appears to have run on exchange between coastal fishing communities and inland farmers growing cotton for nets, alongside squash and beans.

A society that mobilised thousands of people to raise stone platforms, but never fired a pot or wrote a word, fails several of the standard criteria while obviously being more than a collection of villages. Mesoamerica followed later and differently: the Olmec from roughly 1500 BC, and the Maya developing a fully decipherable script in the first millennium BC.

Göbekli Tepe and the problem it causes

In 1995, the German archaeologist Klaus Schmidt began excavating a mound in south-eastern Turkey, near Şanlıurfa, that had been surveyed decades earlier and written off as a medieval cemetery. What came out of it rearranged the timeline.

Göbekli Tepe's oldest enclosures date to around 9500 BC — roughly six thousand years before Sumerian writing, and about seven thousand before Stonehenge. They consist of rings of T-shaped limestone pillars, the largest over five metres tall and weighing in the region of ten tonnes, carved in relief with foxes, boars, snakes, scorpions, vultures and cranes. Quarrying, moving and raising them required organised labour on a scale nobody expected from that period.

The builders were not farmers in the developed sense. Domesticated crops and animals do not appear in the region until centuries later, and the remains at the site point to wild game and wild cereals — though these are the hills where einkorn wheat was eventually domesticated, so "hunter-gatherers" understates a population already managing wild plants intensively.

The standard model had monuments arriving late: farming produces surplus, surplus supports cities, cities build temples. Göbekli Tepe inverts the order. Whether it inverts the causation — Schmidt's suggestion that the demands of feeding the builders helped drive domestication — is much debated and not established. The reading has shifted since his death in 2014, too: he took the site for a purely ritual sanctuary with no residents, while later excavation has found cisterns and domestic debris, and related sites such as Karahan Tepe are still being uncovered nearby.

None of this makes Göbekli Tepe a civilisation on the usual criteria: no city, no writing, no obvious state. What it does is demolish the assumption that monumental building requires those things first.

So which is oldest?

The honest summing up is a list by criterion, not a single winner:

• Writing, cities and state administration together: Sumer, with urban settlement well before 4000 BC and proto-cuneiform by around 3400 BC.
• Oldest unified territorial state: Egypt, from around 3100 BC — and within a century or two of Sumer on writing.
• Largest, most systematically planned early cities: the Indus Valley, mature phase roughly 2600–1900 BC.
• Oldest monumental architecture: Göbekli Tepe, from around 9500 BC, built before farming.
• Oldest permanent settlements: Jericho, with a stone tower from around 8000 BC, and Çatalhöyük from roughly 7100 BC.
• Oldest civilisation in the Americas: Norte Chico in Peru, monumental from around 2600 BC, without pottery or writing.
• Longest continuously used writing system: Chinese, from the oracle bones of around 1250 BC to the present.

Sumer is the defensible answer to the question as usually asked. It is not the answer to every version of it, and the versions it fails are the ones worth thinking about.

Frequently Asked Questions

What is the oldest civilization in the world?

Sumer, in southern Mesopotamia, is the standard answer: it is the earliest known society combining cities, writing, specialised labour and state administration, with settlements from around 5400 BC and written records by roughly 3400 BC. Egypt follows closely and is sometimes given equal billing. The ranking depends entirely on whether writing is treated as a requirement.

Is Sumer or Egypt older?

Sumer is generally considered slightly older, because urban settlement and written records appear there first — proto-cuneiform from around 3400 BC against Egyptian hieroglyphic labels from around 3250–3200 BC. Egypt achieved large-scale political unification around 3100 BC, arguably earlier than Mesopotamia did, so the two swap places depending on the criterion. The gap is small enough that revised dating could close it.

What is the oldest continuous civilisation?

China has the strongest claim, with a writing system running from the Shang oracle bones of around 1250 BC to today — though that span includes conquests, fragmentation and foreign dynasties. Egypt and India are the other usual candidates: Egypt's pharaonic state lasted roughly three thousand years but was absorbed by Persia, Greece and Rome, and its ancient language and religion did not survive. "Continuity" in these debates almost always means one chosen thread — a script, a religion, a territory — rather than an unchanged society.

What is the oldest city in the world?

Jericho, in the West Bank, has the oldest known permanent settlement, with a stone tower dated to around 8000 BC and occupation running back earlier still. Damascus and Aleppo are the most commonly cited "oldest continuously inhabited" cities, with Byblos, Plovdiv and Varanasi also making the claim. All of these are contested, because "continuously inhabited" is hard to prove — occupation gaps are easy to miss, continuous use of a site is not the same as continuity of a city, and digging is limited by the modern city sitting on top.

Keep going

The deeper you push into early history, the more the neat categories give way — and that is usually where the good material is. 10 fascinating history stories collects episodes with the same quality of surprise as Göbekli Tepe, and the best history books for beginners points to writers who handle this material well at book length, including several on Mesopotamia and early states.

If you would rather take it in smaller doses, Chunks publishes one free five-minute narrated story every day — history, science and philosophy — on iOS and Android. It fits the gap where you would otherwise be scrolling: one properly researched story, then done.