Mesopotamian scribes pressed wedge-shaped marks into clay tablets with tools such as reed styluses. The shapes of the writing gradually changed with the tools and the needs of record keeping.

Cuneiform was used for several languages, rather than belonging exclusively to one people. Accounts, contracts, literature, and royal inscriptions could all be recorded with the system.

Proto-cuneiform (c. 3300 BC)

The cuneiform script was developed from pictographic proto-writing in the late 4th millennium BC, stemming from the Near Eastern token system used for accounting. The meaning and usage of these tokens is still a matter of debate. These tokens were in use from the 9th millennium BC and remained in occasional use into the late 2nd millennium BC. Early tokens with pictographic shapes of animals, associated with numbers, were discovered in Tell Brak, and date to the mid-4th millennium BC. It has been suggested that the token shapes were the original basis for some of the Sumerian pictographs.

Mesopotamia's "proto-literate" period spans roughly the 35th to 32nd centuries BC. The first unequivocal written documents start with the Uruk IV period, from c. 3300 BC, followed by tablets found in Uruk III, Jemdet Nasr, Early Dynastic I Ur and Susa (in Proto-Elamite) dating to the period until c. 2900 BC.

Originally, pictographs were either drawn on clay tablets in vertical columns with a sharpened reed stylus or incised in stone. This early style lacked the characteristic wedge shape of the strokes. Most Proto-Cuneiform records from this period were of an accounting nature. The proto-cuneiform sign list has grown, as new texts are discovered, and shrunk, as variant signs are combined. The current sign list is 705 elements long with 42 being numeric and four considered pre-proto-Elamite.

Certain signs to indicate names of gods, countries, cities, vessels, birds, trees, etc., are known as determinatives and were the Sumerian signs of the terms in question, added as a guide for the reader. Proper names continued to be usually written in purely "logographic" fashion.

Archaic cuneiform (c. 2900 BC)

The first inscribed tablets were purely pictographic, which makes it technically difficult to determine which language they represent. Different languages have been proposed, though usually Sumerian is assumed. Later tablets dating after c. 2900 BC start to use syllabic elements, which clearly show a language structure typical of the agglutinative Sumerian language. The first tablets using syllabic elements date to the Early Dynastic I–II periods c. 2800 BC, and they are agreed to be clearly in Sumerian.

About 2800 BC some pictographic elements started to be used for their phonetic syllabic value, permitting the recording of abstract ideas and personal names. Many pictographs began to lose their original function, and a given sign could have various meanings depending on context. The sign inventory was reduced from some 1,500 signs to some 600 signs, and writing became increasingly phonological. Determinative signs were re-introduced to avoid ambiguity. Cuneiform writing proper thus arises from the more primitive system of pictographs at about this time, which historians label the Early Bronze Age II epoch.

The earliest known Sumerian king whose name appears on contemporary cuneiform tablets is Enmebaragesi of Kish (fl. c. 2600 BC). Surviving records became less fragmentary for following reigns, and by the arrival of Sargon it had become standard practice for each major city-state to date documents by year-names, commemorating the exploits of its king.

Sumero-Akkadian cuneiform

The archaic cuneiform script was adopted by the Akkadian Empire from the 24th century BC. The Akkadian language being East Semitic, its structure was completely different from Sumerian. The Akkadians found a practical solution in writing their language phonetically, using the corresponding Sumerian phonetic signs. Still, many of the Sumerian characters were retained for their logographic value as well: for example the character for "sheep" was retained, but was now pronounced immerum, rather than the Sumerian udu. Such retained individual signs or, sometimes, entire sign combinations with logographic value are known as Sumerograms, a type of heterogram.

The East Semitic languages employed equivalents for many signs that were distorted or abbreviated to represent new values because the syllabic nature of the script as refined by the Sumerians was not intuitive to Semitic speakers. From the beginning of the Middle Bronze Age (20th century BC), the script evolved to accommodate the various dialects of Akkadian: Old Akkadian, Babylonian, and Assyrian. At this stage, the former pictograms were reduced to a high level of abstraction, and were composed of only five basic wedge shapes: horizontal, vertical, two diagonals and the Winkelhaken impressed vertically by the tip of the stylus.