The whole of the surviving text comes from a small parchment codex found by Urbain Bouriant of the French Archaeological Mission at Cairo, during excavation of a cemetery at Akhmim in Upper Egypt in the winter of 1886-87. The codex was buried with a monk and is dated on palaeographic grounds to the eighth or ninth century, six or seven hundred years after the text it carries was written. It also contains portions of 1 Enoch and of the Apocalypse of Peter. Bouriant published it in 1892.
Two Oxyrhynchus fragments, P.Oxy. 2949 and P.Oxy. 4009, have been proposed as much earlier witnesses, and the so-called Fayyum fragment has also been suggested. None of these identifications is secure. P.Oxy. 2949 overlaps the Akhmim text loosely rather than exactly, and the case for P.Oxy. 4009 rests on reconstruction. Anyone who states flatly that second-century copies of this gospel survive is overstating the evidence.
Most scholars place composition in the second century, and a substantial number in its first half, on the grounds that the narrative shows knowledge of material found in all four canonical gospels while treating it freely, and that its anti-Jewish framing and its legendary expansion of the guard at the tomb fit that period. John Dominic Crossan argued the opposite direction: that embedded in the text is an independent early source he called the Cross Gospel, written before Mark and used by all four evangelists, so that the canonical passion narratives derive from it rather than it from them. The thesis is set out at length and has been influential in shaping the debate, but it has not been widely accepted; Raymond E. Brown, among others, argued in detail that the text instead reflects a writer working from memory of the canonical gospels, with the errors and conflations that implies. The minority position deserves to be known, and it should not be reported as the consensus.