Question · 20 min read
Are the hadith reliable?
On this page
It depends on the report and the standard being asked to carry it. No surviving hadith document comes from Muhammad’s lifetime, and a substantial revisionist line reads common-link patterns and attribution growth as evidence that much of the corpus took shape in the second and third Islamic centuries. Other historians use isnad-cum-matn analysis, early written transmission, and collector models to date particular material earlier, sometimes to the Companion or Successor generations and potentially beyond. The field does not share one threshold that yields the blanket verdict that no hadith can be traced to Muhammad. The defensible finding is more limited: traditional grades establish reliability within the classical system but do not by themselves establish prophetic origin historically; confidence must be argued report by report, and many central attributions remain disputed.
What the traditional sciences actually check
Reported. Hadith are transmitted with an isnad — a chain of named narrators running from the compiler back through successive generations to a Companion of Muhammad, and thence to the Prophet himself. The classical sciences (mustalah al-hadith) built a sophisticated apparatus for evaluating these chains: assessing each narrator’s reliability and character (jarh wa ta’dil), verifying that the chain is unbroken, and checking whether the transmitters could plausibly have met.
This apparatus deserves the respect it is usually denied in polemic: it is systematic, documented, and within its own premises rigorous. But it has a structural blind spot: the apparatus evaluates narrators individually, and where it registered the shape of a chain (the gharib classification the tradition itself applied to solitary transmission), it treated that shape as a feature to record, not as evidence about when the material entered circulation. The dating inference from the network’s shape is the modern step. A chain of individually reputable narrators can still carry material that entered the stream long after the Prophet, because honest transmission of drifted material produces exactly the same clean chain as honest transmission of authentic material: sincerity at every link is compatible with a false attribution at the origin. That network-level question, where the chains actually converge and when, is where modern critical scholarship has concentrated. The full analysis is set out under “The Transmission Problem” in The Historical Formation of the Hadith Corpus.
When were hadith written down?
Reported. No hadith survives in a document contemporary with Muhammad, and none is independently corroborated by non-Islamic sources of the period. The canonical collections themselves were compiled in the third century of the Islamic era; whether the material in them took shape then or substantially earlier is the disputed question this page tracks.
The customary explanation is an oral culture in which written records were not the norm. The difficulty is that this sits uneasily with the tradition’s own account of the period. Primary. That account describes writing in active use for matters considered important: Quranic revelation was recorded in the Prophet’s lifetime, and the first collection drew on parchment, bone, and palm stalks;1
Sahih al-Bukhari, Kitab al-Tafsir, narrated Zaid bin Thabit; tr. Muhsin Khan (full text). Abu Bakr commissioning the first collection:
“You are a wise young man and we do not suspect you (of telling lies or of forgetfulness): and you used to write the Divine Inspiration for Allah’s Messenger. Therefore, look for the Qur’an and collect it (in one manuscript). […] So I started locating Qur’anic material and collecting it from parchments, scapula, leaf-stalks of date palms and from the memories of men (who knew it by heart).”
the Treaty of Hudaybiyyah was drafted in writing, with a dispute conducted over its exact wording;2
Sahih al-Bukhari, Kitab ash-Shurut, narrated al-Miswar bin Makhrama and Marwan; tr. Muhsin Khan (full text).
“So, the Prophet called the clerk and said to him, “Write: By the Name of Allah, the most Beneficent, the most Merciful.” Suhail said, “As for ‘Beneficent,’ by Allah, I do not know what it means. So write: By Your Name O Allah, as you used to write previously.” […] Then he dictated, “This is the peace treaty which Muhammad, Allah’s Messenger has concluded.” Suhail said, “By Allah, if we knew that you are Allah’s Messenger we would not prevent you from visiting the Kaba, and would not fight with you. So, write: "Muhammad bin Abdullah.""
the Constitution of Medina is presented as a written instrument; written correspondence with foreign rulers is reported; and Mecca is portrayed as a commercial centre where written contracts were in use.
Argued (historical-critical). On the tradition’s own account, writing was available and applied to important material, while no contemporary documentary corpus of the sayings later elevated to the second source of law is known. That is an inference from the surviving record, not proof that no such record ever existed. Three readings are open.
The first reading: the sayings were not, at the time, regarded as authoritative in the way later doctrine holds them to be — which weakens the theological status subsequently assigned to hadith. The second: the sayings did not yet exist in the form the tradition presents, which supports the reconstruction thesis directly.
The third reading is the tradition’s own answer. Primary. Sahih Muslim, Kitab al-Zuhd, carries the Prophet’s instruction not to write down anything from him other than the Quran, to erase what had been written, and to narrate from him orally.3 Sahih Muslim, Kitab al-Zuhd wa’l-Raqa’iq, narrated Abu Sa’id al-Khudri; tr. Abdul Hamid Siddiqui (full text). An aside by the transmitter Hammam, recalling that the word “deliberately” was also said, is trimmed: “Do not take down anything from me, and he who took down anything from me except the Qur’an, he should efface that and narrate from me, for there is no harm in it and he who attributed any falsehood to me […] he should in fact find his abode in the Hell-Fire.” On this reading the absence of a contemporary record is not silence but obedience: the sayings were authoritative, while writing was withheld to protect the Quran’s exclusivity. The tradition also carries permissions to write: Abdullah ibn Amr reportedly wrote what he heard with the Prophet’s approval, and the Prophet reportedly ordered a sermon written for Abu Shah. Umar’s fear of producing a book beside the Book appears within the same body of accounts. The tradition reconciles prohibition and permission by period or person.
The third reading rests on reports that are themselves hadith, the corpus explaining its own origins through the channel in question, so the explanation is drawn from the very body of material whose reliability is in question. It also coexists with contrary permission reports, which the harmonizations accommodate rather than dissolve. A deliberate policy of non-writing that was repeatedly waived therefore explains only part of the missing documentary record.
Gregor Schoeler’s work complicates this argument. He argues for hybrid oral-written transmission in which written notes served as memory aids inside a fundamentally oral system—a mode that would not be expected to leave surviving documents. This weakens the argument from silence without dissolving it: private memory aids still do not amount to the contemporaneous authoritative record that the theological claims about hadith would lead one to expect.
One point stands independently of how these readings are weighed: the entire edifice of isnad criticism exists because direct documentary evidence is absent. Where contemporary records survive, chain-of-narrator authentication is unnecessary. The chains are what there is — which is why their structure carries the weight of the question. The documentary record, and the three readings it leaves open, are treated at depth under “The Absence of Contemporary Documentation” in The Historical Formation of the Hadith Corpus.
What the chains look like when they are mapped
Reported. When all recorded isnads for a given hadith are mapped as a network, a recurring topology emerges. Rather than radiating outward from the Prophet through many independent early strands, the chains frequently pass through a single transmitter living generations — often around a century — after Muhammad. Below this figure — the common link — the chains are thin, often a single strand; above him, they branch widely, in the pattern expected of genuine circulation. Joseph Schacht identified the pattern in The Origins of Muhammadan Jurisprudence (1950); G.H.A. Juynboll developed the analysis substantially in the decades after.
Their inference is that the common link marks the point at which a tradition entered circulation, and that the chain reaching back beyond him is a later construction supplying prophetic authority to material that originated with, or shortly before, the common link himself. Traditional hadith criticism cannot detect this, because the pattern is a property of the network, not of any narrator in it.
The tradition’s own taxonomy contains the direct counter-claim. Reported. The classical sciences classify as mutawatir a report transmitted at every generation by numbers large enough, on the tradition’s account, to preclude coordinated fabrication, a report that therefore yields certain knowledge; below it sits mashhur, widespread transmission that falls short of tawatur at the first generation. The warning against lying about the Prophet quoted later on this page is the tradition’s stock example: it is reported as claimed tawatur from dozens of Companions. If those claims hold, the common-link topology is not the whole picture: there exists a class of reports whose early transmission is anything but a single strand. The critical reply is Juynboll’s, and it is presented as his position rather than a neutral finding: he argued that claimed tawatur dissolves under network analysis, that when the parallel chains are actually mapped, they converge on later figures rather than running independently back to the Companions named. The reader has both claims: a category built to certify mass transmission, and an analysis contending that the mass is later than it looks.
A worked example
Primary. The hadith innama al-a’mal bi’l-niyyat — “actions are judged by their intentions” — opens Sahih al-Bukhari and is among the most widely cited in the corpus. The tradition itself classifies it as gharib — solitary transmission: here, a single strand through four successive narrators. That chain runs Umar ibn al-Khattab → Alqama ibn Waqqas → Muhammad ibn Ibrahim al-Taymi → Yahya ibn Sa’id al-Ansari, who died around 143 AH — and at Yahya, transmission expands abruptly to a large number of narrators.4 Sahih al-Bukhari, Kitab Bad’ al-Wahy, narrated Umar bin al-Khattab; tr. Muhsin Khan (full text). The Arabic opens innama al-a’mal bi’l-niyyat, “actions are only by intentions”; “The reward of” is the translator’s gloss. On this rendering, see Is the English translation of the Quran accurate?. The single early chain (Umar → Alqama ibn Waqqas → Muhammad ibn Ibrahim al-Taymi → Yahya ibn Sa’id al-Ansari) is the early portion recorded in Bukhari’s isnad for this hadith: “I heard Allah’s Messenger saying, “The reward of deeds depends upon the intentions and every person will get the reward according to what he has intended. So whoever emigrated for worldly benefits or for a woman to marry, his emigration was for what he emigrated for."" The singularity of the early chain is not a revisionist discovery; classical scholars observed it and discussed it under the category of gharib.
The structural reading takes the observation one step further: a saying of this significance transmitted by exactly one person per generation for roughly 130 years, then suddenly by dozens, is more economically explained by origination at the point of expansion than by an improbably narrow bottleneck before it.
That inference is probabilistic, not demonstrative, and the rival explanation carries real weight. A saying may genuinely have originated with the Prophet, been heard by few, passed quietly through a narrow line, and only later acquired the currency that produced wide transmission. Nothing in the isnad structure alone excludes this. The revisionist case rests on two claims together: that such narrow survival is improbable for material of this importance, and that the pattern recurs too systematically across the corpus to be repeated coincidence. That is a strong argument from the shape of the evidence rather than a proof, and because this hadith is the standard illustration of the thesis, it is also the case against which the tradition’s defenders have written most directly.
The counter-scholarship
The common-link thesis is the leading revisionist account, not an unchallenged one.
One line of response rests on a check that can be put in ordinary words: do the different chains that carry a saying also carry genuinely different wordings of it (a sign of real independent transmission), or one wording, as if every chain led back to a single source?
Reported. Harald Motzki’s isnad-cum-matn method examines chains and texts together, asking whether textual variants across branches correlate with the points where the chains diverge. Where they do, that is evidence of genuinely independent transmission rather than one fabricator generating multiple chains. Applying this to material in Abd al-Razzaq’s Musannaf and elsewhere, Motzki argued that certain traditions can be dated earlier than their apparent common links. Gregor Schoeler developed related arguments about the interplay of oral and written transmission in the early period, suggesting the absence of early written collections is less anomalous than Schacht supposed. From within the tradition, M.M. Azami mounted a sustained critique of Schacht’s chronology and use of sources.
Motzki challenges the common-link inference itself. On his reading, the common link may be the first systematic collector rather than the report’s originator: an early specialist gathered material and taught it professionally, causing recorded transmission to widen at that point. The single strands below the common link may reflect a collector naming a principal source rather than every route by which the material travelled.
The network’s shape therefore cannot by itself distinguish a fabricator from a collector. Motzki’s claim is underdetermination, not proof of authenticity: the topology does not establish that the lower strands are false.
Two features of this counter-scholarship qualify it. Its successes often push dating back to the Companion or senior Successor generation without automatically establishing prophetic wording; the gap narrows and must then be assessed case by case. The common-link pattern, for its part, remains an important structural description; what the collector reading contests is what the pattern licenses one to infer. Between origination, dissemination, and first systematic collection the topology alone underdetermines the answer. Preferring Juynboll’s reading for a particular bundle is this site’s argued historical judgment, not the field’s settled economy.
Both are judgments rather than neutral findings, and this page presents them as such. A scholar persuaded by Motzki would arrange the same material to different effect, in good faith.
The dispute has not stood still, and the newer work cuts both ways. Andreas Görke and Gregor Schoeler applied the isnad-cum-matn method to the corpus of ‘Urwa ibn al-Zubayr and argued that a core of accounts was being taught in Medina in the late first Islamic century, a dating Stephen Shoemaker contested at precisely the reach back to ‘Urwa, in a published exchange that remains open. Joshua Little applied the same method to the reports of Aisha’s marital age and reached a skeptical result, an archetype formulated in the mid-second century, in an examined Oxford doctoral thesis. That case shows at minimum that the method does not mechanically guarantee early dating. And Sean Anthony’s work on the earliest non-Islamic sources sets an external floor (Muhammad attested as a prophet-figure within roughly a decade of his death) that corroborates the man and his period, not the content of the hadith corpus. The post-2010 field is surveyed, with each finding at its verified strength, under “The Field Since 2010” in The Historical Formation of the Hadith Corpus.
The direction of attribution
Reported. A separate line of argument tracks how attributions changed over time. Schacht observed that comparing legal texts of different dates shows the same substantive ruling attributed to progressively more authoritative figures: a ruling appearing in early Kufan material as the opinion of the local jurist Ibrahim al-Nakha’i (d. c. 96 AH) appears in later material as the opinion of Ibn Mas’ud, the Companion associated with Kufa — and later still as a prophetic hadith with a complete chain. Malik’s Muwatta shows a comparable pattern internally: rulings Malik presents simply as the settled practice of Medina, without prophetic attribution, appear in the later canonical collections as prophetic statements furnished with full isnads. The comparative evidence is set out under “The Direction of Attribution” in The Historical Formation of the Hadith Corpus.
In the documentable cases, the observed movement runs consistently toward higher authority, and the reverse movement is not a notable feature of the material. That is what one would expect if the operative mechanism was competitive legitimation rather than progressive recovery of better sources. It should be read as a strong tendency identified across Schacht’s survey and its successors — not as an established universal.
The Medinan cases permit a further reading. Continuous communal practice—the channel the prayer page treats as firmer than any isnad—is exactly what Malik’s rulings without isnads claim as their basis: the settled public practice of Medina across generations. A reading associated with M.M. Azami and, on Malik specifically, reported from Yasin Dutton’s work, holds that where a ruling genuinely rested on such practice, prophetic chains acquired later may be formalisation rather than invention. The content could have been transmitted as practice before receiving the documentary apparatus that a competitive legal culture came to demand.
On that reading, the Muwatta pattern records a change in the form of authority rather than the manufacture of its content. The escalation argument remains strongest where the relative chronology of the compared texts is independently secure and no practice channel plausibly carries the ruling. The ʿamal cases sit at its boundary.
The argument depends on the relative dating of the compared texts. Where chronology is independently secure, as with the Kufan material, the inference is strong. Where a source’s date is inferred from its contents, the reasoning risks circularity.
The inheritance maxim la wasiyya li-warith (“there is no bequest to an heir”) illustrates the limit. It reaches the canonical collections as a pronouncement during the Farewell Sermon, while juristic literature also argues for the rule on its own terms. Deciding whether this is documented escalation or a genuinely early rule requires independently established chronology in each case.
Depth: how a corpus gets built without a conspiracy
The escalation dynamic does not require bad faith to run. The garrison towns of the expanding empire — Kufa, Basra, Fustat, Damascus, alongside Medina — faced immediate administrative demands: taxation, land settlement, conquest revenue, marriage across newly mixed populations, criminal jurisdiction. Local legal practice (amal) developed to meet them, resting at first on the authority of local scholars and governors. As regional schools consolidated and came into competition, local authority became insufficient to settle disputes between them, and the available escalation ran from local jurist to Companion to Prophet. Once any school grounded its positions in prophetic hadith, the others faced pressure to match. On this account the fabrication was not primarily conspiratorial but institutional: each escalation locally rational, the cumulative outcome a prophetic corpus constructed to adjudicate second- and third-century legal questions.
Whether the participants understood themselves to be fabricating is a separate question from what was produced. Primary. The corpus itself preserves the warning that whoever deliberately lies about the Prophet has secured his place in the Fire,5 Sahih al-Bukhari, Kitab al-‘Ilm, narrated Abu Huraira; tr. Muhsin Khan (full text). The parenthetical “(intentionally)” renders the Arabic muta’ammidan, “deliberately,” which is present in the text: “And whoever tells a lie against me (intentionally), then (surely) let him occupy his seat in Hell-fire.” and the tradition built an entire science of forged traditions (mawdu’at) — demonstrating that deliberate fabrication was recognised as widespread and understood as a grave offence. Attributing specific words to Muhammad on no evidence that he spoke them falls within the prohibition the participants themselves acknowledged.
The Quran is a different case
Reported. The Quran occupies a different and substantially stronger evidentiary position than the hadith corpus, and the two should never be assimilated. Early manuscripts — the Birmingham folios, the Sana’a palimpsest, the Codex Parisino-petropolitanus — support the circulation of Quranic material in written form within the first Islamic century, close to the traditional chronology. The Quran is not a second-century composition in the way that much of the hadith corpus appears to be; a critic who extends the hadith argument to the Quran is overreaching, and the tradition is right to object. What the manuscript record does and does not establish — early text, human standardization, alternatives destroyed — is treated at Was the Quran perfectly preserved? and What does the Sana’a manuscript show?, and in the reference document under “The Quranic Text” in The Historical Formation of the Hadith Corpus.
Verdict
The scope of the problem depends on how much of Islamic practice rests on hadith rather than on the Quran — and the answer is: most of it. Reported. The Quranic text does not specify the form or number of the daily prayers, the rates and thresholds of zakat, most of the mechanics of inheritance beyond the fixed shares, the detailed procedures of marriage and divorce, the prescribed punishments for several offences including apostasy, or the greater part of ritual and dietary law as practised. All of that derives from hadith and the juristic tradition built on it. The prayer case is treated at Why isn’t prayer described in the Quran?; two worked examples of law resting on hadith are at Stoning is not in the Quran — so why is it Islamic law? and What is the punishment for leaving Islam?.
If the revisionist historical-critical account is broadly correct, the consequence is not that Islam lacks a foundation. The Quran remains, and its early dating is well supported. The consequence is that much of the practised religion — its law, ritual forms, and daily obligations — rests on attributions whose prophetic origin remains historically disputed and whose recorded forms often become visible only in the second and third Islamic centuries. The stronger claim that the whole practised religion is historically untraceable to Muhammad exceeds what the surveyed dispute establishes.
The common-link analysis identifies a real and widespread feature of the isnad literature whose interpretation remains debated. The reviewed evidence does not justify a universal negative about every hadith under a single set of “ordinary historical standards.” It does establish the relevant limit: a traditional grade does not itself demonstrate prophetic origin by historical-critical methods, and the authority assigned to a report must not be confused with historical certainty about its attribution.
Sources on this page: Sahih al-Bukhari, Kitab Bad’ al-Wahy (the intentions hadith and its chain), Kitab al-‘Ilm (the lie-against-me warning), Kitab al-Tafsir (Zayd’s collection account), Kitab ash-Shurut (the Hudaybiyyah treaty); Sahih Muslim, Kitab al-Zuhd (the writing prohibition). Scholarship cited in the body: Joseph Schacht, The Origins of Muhammadan Jurisprudence (1950); G.H.A. Juynboll, Muslim Tradition (1983) and Encyclopedia of Canonical Hadith (2007); Harald Motzki, The Origins of Islamic Jurisprudence (2002), “Dating Muslim Traditions: A Survey” (Arabica 52, 2005), and Analysing Muslim Traditions (2010, with Nicolet Boekhoff-van der Voort and Sean W. Anthony); Gregor Schoeler, The Oral and the Written in Early Islam (2006); Andreas Görke & Gregor Schoeler, Die ältesten Berichte über das Leben Muhammads (2008), with Stephen Shoemaker’s response in Der Islam 85 (2011) and the joint debate in Der Islam 89 (2012); Joshua Little’s Oxford doctoral thesis on the Aisha marital-age hadith (2022); Sean W. Anthony, Muhammad and the Empires of Faith (2020); M.M. Azami, On Schacht’s Origins of Muhammadan Jurisprudence (1985); Yasin Dutton’s work on Malik and Medinan practice (cited at reported strength). Full treatment: The Historical Formation of the Hadith Corpus · How this site works