Hadith Sciences — How Hadith Are Graded

🕰️ At first there was no need of it

In the Prophet's ﷺ lifetime and the years just after, verification was face to face. The Companions heard from him directly, corrected one another, and could walk across Madinah to ask. Transmission was already a duty; he ﷺ commanded it in the same breath as he warned against corrupting it:

Transmit — but transmit truthfully. Those two commands arrive together, and the entire science that followed is simply the second one taken seriously across centuries.

⚡ Then the fitnah, and the question changed

After the murder of ʿUthmān (may Allah be pleased with him) the community fractured politically, and parties appeared with positions to defend. Some began putting words into the Prophet's ﷺ mouth to serve their side. He ﷺ had foretold exactly this:

The response of the scholars is preserved in the introduction to Ṣaḥīḥ Muslim, which is itself a small textbook of this science. Muḥammad ibn Sīrīn, the great Tābiʿī of Basra, described the exact moment the discipline was born:

Three sentences from the same introduction supply the whole rationale. Ibn Sīrīn again: "Indeed this knowledge is faith, so carefully consider from whom you take your faith" (Ṣaḥīḥ Muslim — site reader no. 7393). Saʿd ibn Ibrāhīm: "There is to be no narrating on authority of the Messenger of Allah, may Allah send blessings and peace upon him, except by trustworthy narrators (Thiqāt)" (Ṣaḥīḥ Muslim — site reader no. 7398). And ʿAbdullāh ibn al-Mubārak, who put the case in one line that has been quoted ever since: "The chain of narration is from the Dīn, and were it not for the chain of narration whoever wished could say what he wanted" (Ṣaḥīḥ Muslim — site reader no. 7399).

The Quran had already laid down the principle of checking a report before acting on it: "O ye who believe! If a wicked person comes to you with any news, ascertain the truth, lest ye harm people unwittingly, and afterwards become full of repentance for what ye have done" (Sūrat al-Ḥujurāt 49:6). If an ordinary piece of news from an unreliable carrier must be verified before you act, how much more a report that will be prayed by, fasted by, married and inherited by?

Refs: Ṣaḥīḥ Muslim — site reader nos. 7383, 7393, 7394, 7398, 7399 (Imam Muslim's introduction); Quran 49:6. See also how hadith were collected and authenticated.

✍️ Writing was there from the start

Abū Hurayrah (may Allah be pleased with him), the most prolific narrator of all, named the one Companion who out-recorded him, and the reason:

And the Prophet ﷺ ordered a sermon written out on request. After the conquest of Makkah he delivered the ruling on its sanctity, and a Yemeni man in the crowd asked for a copy:

What the later centuries added was not writing but systematic collection, arrangement and criticism — the gathering of scattered notebooks and memorised holdings into organised, chain-verified books. See when hadith were written down and the role of the Sahābah in transmitting hadith.

🎯 Exact wording mattered to him ﷺ

Al-Barāʾ ibn ʿĀzib was taught a duʿāʾ for bedtime and recited it back to the Prophet ﷺ to check it. He substituted one synonym — rasūl for nabī, messenger for prophet, both true of him ﷺ — and was corrected on the spot:

If a single word in a duʿāʾ was worth correcting, the standard for reporting his speech generally is clear. The same care appears in his ﷺ supplication for the accurate transmitter:

As he heard it — verbatim fidelity is a virtue the Prophet ﷺ prayed for by name. That duʿāʾ is the reason this site quotes hadith word for word from its corpora and never paraphrases into quotation marks. Where wordings of one event genuinely differ between narrators, the scholars discuss riwāyah biʾl-maʿnā, narration by meaning, and its strict conditions — see why hadith wordings differ.

🔎 They checked each other, and they demanded witnesses

The Companions were upright, and Ahl as-Sunnah hold them collectively trustworthy in what they report from the Prophet ﷺ — he ﷺ testified to their standing: "The best people are those of my generation, and then those who will come after them (the next generation), and then those who will come after them" (Ṣaḥīḥ al-Bukhārī 6429). Uprightness, however, is not the same thing as infallible memory, and they knew it. ʿĀʾishah (may Allah be pleased with her) corrected Ibn ʿUmar's count of the Prophet's ﷺ ʿumrahs from inside her dwelling, and note her wording — a prayer for him, not an accusation:

That single exchange contains the seed of the deepest branch of the science: an honest, upright narrator can still be mistaken, so reports are compared against reports. ʿUmar ibn al-Khaṭṭāb (may Allah be pleased with him) went further and required corroboration for a report he had not himself heard. Abū Mūsā al-Ashʿarī had left ʿUmar's door after seeking permission three times, citing the Sunnah for it:

ʿUmar did not doubt Abū Mūsā's honesty; he was establishing a standard of care in the community while the Prophet's ﷺ generation was still alive. The same incident is recorded in Ṣaḥīḥ Muslim — site reader no. 5480. Corroboration, cross-examination, and the naming of sources were therefore not later inventions imposed on the material. They began inside it.

🔗 Isnād — the chain

The isnād (also called the sanad) is the chain of transmitters: A narrated to us, from B, from C, from the Prophet ﷺ. It is the narration's provenance record, and the critics interrogated every link with three questions:

• ʿAdālah — was this narrator upright: Muslim, of sound religion, not known for lying or major sin, and not reckless in what he passed on?
• Ḍabṭ — was he precise: a reliable memory, or a carefully kept and guarded written record? A saint with a poor memory does not qualify.
• Ittiṣāl — is the chain actually joined: did each man really meet and hear from the one above him, with no silent gap?

To answer those questions at scale, the scholars built the science of rijāl — biographical dictionaries recording when each narrator was born and died, in which cities he lived, whom he studied under, whom he taught, and what the critics said about his reliability. Tens of thousands of transmitters are catalogued in this literature. It is the apparatus that makes a grading checkable rather than a matter of opinion.

Matn — the text

The matn is the wording itself: what the Prophet ﷺ said, did, or silently approved. It is examined too. Does this wording contradict the decisive text of the Quran? Does it contradict what stronger narrators transmit from the same teacher? Does it contain something impossible, or a promise of reward so wildly out of proportion to the deed that forgery is the likeliest explanation? A clean-looking chain carrying an anomalous text is not waved through — a narration that fails on the matn is called shādh (anomalous) or is said to contain an ʿillah (a hidden defect). Both halves must stand. See the matn and textual authenticity and what a shādh narration is.

Related: what is a hadith and how is it authenticated · what makes a hadith ṣaḥīḥ

🧾 Naming an unreliable narrator is advice, not slander

The question was put directly to the greatest jurists of the second century, and Yaḥyā ibn Saʿīd al-Qaṭṭān recorded their unanimous answer:

The principle is the one that governs all sincere warning: where silence would let a Muslim be harmed — here, harmed in his religion — naming the defect is a duty, not a sin. The same scholars were scrupulous about scope. They criticised a man's transmission, not his person, and only to the extent needed.

Ṭāwūs gave the layman's version of the rule when a narration was brought to him by an unnamed source: "If your companion is trustworthy, then take from him." (Ṣaḥīḥ Muslim — site reader no. 7395). Trustworthiness first, content second.

⚖️ Piety alone was never enough

The most counter-intuitive rule in the science is that a man could be a worshipper, an ascetic, admired for his religion, and still be rejected as a transmitter — because sincerity does not confer accuracy, and enthusiasm for the good has its own failure mode. Yaḥyā ibn Saʿīd al-Qaṭṭān put it bluntly:

He was describing men who invented or loosely repeated narrations urging prayer, fasting and charity, believing they served religion by it. The critics refused the trade. The converse held equally, in Ayyūb as-Sakhtiyānī's question about a man of doubtful religion whose reports were being collected: "Have you seen a man whose Dīn you do not trust- how do you trust him regarding Ḥadīth?" (Ṣaḥīḥ Muslim — site reader no. 7437). Both conditions — uprightness and precision — had to hold at once, in every link, or the chain failed.

Imam Muslim filled pages of his introduction with concrete verdicts of this kind, naming men the critics had abandoned and why. It is worth knowing that when a modern reader is told a narration is weak "because of so-and-so in its chain", that judgement is drawing on a documented record that begins in books like this one. For the scholars who built and preserved it, see Great Scholars of Islam.

Refs: Ṣaḥīḥ Muslim — site reader nos. 7395, 7402, 7407, 7437 (Imam Muslim's introduction).

⚖️ The four working grades

Ṣaḥīḥ itself has ranks. The strongest tier is what al-Bukhārī and Muslim both record (muttafaqun ʿalayhi); then what one of the two carries alone; then narrations meeting their conditions outside the two books; then the remaining ṣaḥīḥ. This is why, when a narration appears in both Ṣaḥīḥs, the pages of this site cite al-Bukhārī. Below ḍaʿīf the critics also distinguished degrees — ḍaʿīf jiddan (severely weak), munkar (a weak narrator contradicting reliable ones), matrūk (abandoned, its narrator accused of lying) — because a narration with one shaky link is not in the same country as one carried by an accused forger.

Related: ṣaḥīḥ, ḥasan, ḍaʿīf and mawḍūʿ compared · the punishment for fabricating hadith

🚫 Three distinctions readers stumble over

• Ḍaʿīf does not mean proven false. It means the transmission is too weak to establish that the Prophet ﷺ said it. The content might be true in itself; it might even be established by other, sound evidence. What the grade denies is the attribution. We do not put words in his ﷺ mouth on a maybe — which is exactly the warning of Ṣaḥīḥ al-Bukhārī 108 quoted at the top of this page.
• Ḍaʿīf is not a lighter grade of evidence. Rulings of ḥalāl and ḥarām, and every matter of creed, are built on ṣaḥīḥ and ḥasan narrations only. Some scholars permitted acting on a weak narration to encourage a good deed under strict conditions — the deed must already be established by sound evidence, the weakness must not be severe, and one must not assert that the Prophet ﷺ said it. Others held that the authentic Sunnah is abundant enough that nothing needs supplementing from a doubtful source, and that is the safer course this site follows. See can a weak hadith be acted upon?
• Appearing in a hadith book is not a grade. The four Sunan compilers deliberately included material of varying strength so that scholars could examine and rule on it; several of them said so explicitly. Presence in a book is a shelf, not a verdict. The verdict is the assessment of the chain — which is why every Sunan narration in the site's hadith repository carries its grade beside it.

⬆️ Attribution — marfūʿ, mawqūf, maqṭūʿ

• Marfūʿ (raised) — the statement or act is attributed to the Prophet ﷺ himself. This is hadith in the full sense, and only this can be evidence in the way a Sunnah is evidence.
• Mawqūf (stopped) — it stops at a Companion: his own statement, verdict or practice. Valuable, sometimes decisive in fiqh, and occasionally treated as marfūʿ where the subject is one no Companion could have reasoned to on his own. But a Companion's opinion reported as though it were the Prophet's ﷺ word is a serious error.
• Maqṭūʿ (severed) — it stops at a Tābiʿī or later. Useful for knowing how the early generations understood a matter; not a proof from the Prophet ﷺ.

Confusing these three is one of the commonest mistakes in casual quotation, and the difference is set out at marfūʿ versus mawqūf. A related label, ḥadīth qudsī, marks a narration in which the Prophet ﷺ reports words from his Lord in wording that is not Quran — see ḥadīth qudsī compared with the Quran.

🔓 Breaks — mursal, munqaṭiʿ, muʿḍal, muʿallaq

• Mursal — a Tābiʿī narrates directly from the Prophet ﷺ without naming the Companion he heard it from. The missing man is very likely a Companion and therefore trustworthy, which is why the schools differ over accepting it; the majority of hadith critics classify the mursal as a species of weak, since the omitted link cannot be examined. See what a mursal hadith is.
• Munqaṭiʿ — a single link is missing somewhere in the middle of the chain.
• Muʿḍal — two or more consecutive links are missing. Weaker still.
• Muʿallaq (suspended) — one or more links are dropped from the beginning of the chain, at the compiler's own end. Al-Bukhārī uses this deliberately in his chapter headings; where he states it in the active voice, scholars treat it as affirmed by him, and where he states it in the passive, it needs checking elsewhere.

A subtler failure is tadlīs: a narrator who did meet his teacher conceals a gap by using a vague connector such as from rather than I heard, so that a chain with a hole in it looks continuous. The critics identified the men known to do this and required an explicit statement of hearing before accepting their narrations. See tadlīs and the mudallas hadith.

📚 Two masters of one craft

Muḥammad ibn Ismāʿīl al-Bukhārī (d. 256 AH) spent years travelling between the centres of learning of the Muslim world — Makkah, Madinah, Basra, Kufa, Baghdad, Egypt, Syria, Khurasan — gathering narrations from hundreds of teachers and testing every chain against the strictest standard of his generation. He required positive evidence that each narrator had actually met the one he transmits from, not merely that he could have. From the enormous mass he had collected and memorised he selected only what passed, arranging the result by fiqh topic under chapter headings that carry his own legal conclusions. Counting the repeated routes and wordings that he cites in different chapters, the site's Bukhārī corpus holds 7,552 numbered entries; the distinct narrations are far fewer, because a single hadith is often brought several times through different chains, each chain being part of the evidence rather than padding.

His student Muslim ibn al-Ḥajjāj (d. 261 AH) compiled a Ṣaḥīḥ of his own with a complementary method. Rather than scattering a narration across the book, he gathers all its strong routes in one place, wording beside wording, so that a reader can see precisely how the transmissions corroborate one another and where they differ. He also prefixed the work with an introduction on the science itself — the duty of scrutinising narrators, and the verdicts of the critics on named men — and it is from that introduction that the statements of Ibn Sīrīn, Ibn al-Mubārak, Ṭāwūs, Yaḥyā ibn Saʿīd and Ayyūb quoted above are taken. The site's Muslim corpus holds 7,458 entries, that introduction included.

Neither man claimed to have gathered every authentic hadith, and neither said that what he left out was weak. The claim was narrower and stronger: everything included meets the author's stated conditions. The ummah then received both books with acceptance, generation after generation, and centuries of specialists have combed their chains — that sustained scrutiny, which found them sound, is itself part of their standing. On the handful of narrations later critics discussed, and what the discussion actually amounts to, see are all hadith in Bukhārī and Muslim authentic?

Read the collections: Hadith — the full collections · Verified Hadith Collection

🏛️ Al-kutub as-sittah

Six collections became the working backbone of Sunni hadith study:

1. Ṣaḥīḥ al-Bukhārī (d. 256 AH) — authentic-only by the author's condition, arranged by fiqh topic.
2. Ṣaḥīḥ Muslim (d. 261 AH) — authentic-only by the author's condition, arranged to gather the routes of each narration together.
3. Sunan Abī Dāwūd (d. 275 AH) — concentrated on the narrations from which legal rulings are derived.
4. Jāmiʿ at-Tirmidhī (d. 279 AH) — notable for stating a verdict on narrations and recording the fiqh positions built on them, which makes it a teaching book as much as a collection.
5. Sunan an-Nasāʾī (d. 303 AH) — known for exacting narrator criticism and fine attention to defects in chains.
6. Sunan Ibn Mājah (d. 273 AH) — the sixth of the set, carrying the widest range of strength, so that grading matters most of all here.

The crucial difference is the one already stated: the two Ṣaḥīḥs admitted only what their authors held authentic, while the four Sunan contain ṣaḥīḥ, ḥasan and ḍaʿīf material side by side, because they were compiled as working references for scholars rather than certified anthologies. Beyond the six lie many other collections — the Musnad of Aḥmad, the Muwaṭṭaʾ of Mālik, the Sunan of ad-Dāraquṭnī and al-Bayhaqī, the Muʿjams of aṭ-Ṭabarānī — all subject to the same grading. See the six major collections.

🌐 How many independent channels carry it?

Beside grading individual chains, the scholars classified narrations by how many independent routes carry them at each generation:

• Mutawātir — transmitted at every level by so many separate narrators that their agreeing on a lie is inconceivable. The wording of the Quran, the number and form of the daily prayers, the description of wuḍūʾ and of the Ḥajj rites, and the warning against lying on the Prophet ﷺ itself all reach us this way. Mutawātir reports yield certainty.
• Āḥād — everything below that threshold: one, two or a handful of chains at some generation. Most hadith fall here, including most of al-Bukhārī and Muslim. The sub-categories mashhūr, ʿazīz and gharīb describe three, two or one route respectively — see mashhūr, ʿazīz and gharīb.

A point the salaf were firm about: an authentic āḥād report is a proof, in practice and in belief. The Prophet ﷺ sent single men to teach whole regions their religion, and the people were obliged to accept what they taught:

Muʿādh went alone, and what he conveyed bound those who received it — beginning with the testimony of faith itself, which is a matter of creed. Āḥād versus mutawātir therefore describes the breadth of transmission; it is not a licence to set aside an authenticated report because only one reliable route carries it. Once a narration is established as ṣaḥīḥ, it is accepted and acted upon. For the fuller comparison see mutawātir versus āḥād, and on the place of hadith beside the Quran, why hadith are needed to understand the Quran and the answer to those who reject hadith altogether.

🔍 Grading is expert judgement, not arithmetic

Readers sometimes meet the same narration called ṣaḥīḥ by one scholar and ḍaʿīf by another and wonder whether the whole science is guesswork. It is not — but it is ijtihād: trained specialists applying settled criteria to evidence, and honest experts can weigh evidence differently. The usual sources of difference:

• Assessing a narrator. One critic knew a man's early, sound transmission; another met what he narrated after his memory declined, or after he moved to a city where his books were not with him. Both verdicts can be honestly grounded in different slices of the same life.
• Whether the chain truly connects. Did B actually hear this from A? Al-Bukhārī required evidence of an actual meeting; Muslim accepted contemporaneity with the genuine possibility of meeting. Two defensible standards, and they part company on real cases.
• Hidden defects. The science of ʿilal is its deepest layer: a chain that looks flawless but which the comparison of every known route shows to contain an error — a name skipped, a Companion's own words mistakenly raised to the Prophet ﷺ, two narrations fused into one. Critics differ over whether a given oddity is a defect or an acceptable variant.
• How much corroboration is enough. A slightly weak route can be strengthened by parallel routes. Where exactly the support becomes sufficient to lift a narration to ḥasan is itself a judgement.

Note also that a grade attaches to a chain, not to a sentiment: one wording of a narration may be ṣaḥīḥ while a longer wording of the same story, carried by a weaker route, is not. Two scholars can therefore both be right about different routes and appear to contradict each other. Where two soundly authenticated narrations seem to conflict in meaning, the scholars reconcile, or identify which abrogates, or defer — never discard; see when two authentic hadith appear to contradict.

What follows for a reader is modesty in both directions. It is not for the untrained to overturn a grading because a narration displeases him, nor to broadcast a narration as authentic because one website says so. Where this site's sources record a genuine difference on something that matters, the page says so plainly instead of choosing silently.

Related: why hadith numbers differ between editions · The four madhhabs

📵 Repeating without checking is itself a fault

The warning is not about deliberate invention. It is about the man who repeats whatever reaches him, indiscriminately — because some of what reaches him will be false, and he becomes the road it travels on. ʿUmar ibn al-Khaṭṭāb (may Allah be pleased with him) said the same in the same words (Ṣaḥīḥ Muslim — site reader no. 7376), and Imam Mālik added the consequence for anyone hoping to be relied upon: "Know that a man who relates everything he hears is not safe, and he can never be an Imām, as long as he narrates everything he hears" (Ṣaḥīḥ Muslim — site reader no. 7377).

A workable discipline for a reader of this site:

1. Look for the source. A narration circulating with no collection and no number attached has already failed the first test the ummah ever applied to a report.
2. Look up the number. Use the hadith reader or the repository and read the row for yourself. A great deal of what circulates dissolves at this step: the number belongs to something else entirely.
3. Check the grade, not the book. For the four Sunan the repository states a grade for each narration; a narration graded weak is not passed on as the Prophet's ﷺ word.
4. Check the famous ones against the list. Sayings that everyone knows are the likeliest to be baseless — see Popular but Not Authentic, and the well-known China narration as a worked example.
5. When you do not know, say so. I have heard this but I have not verified it costs nothing and breaks the chain of a lie.

See also how to check whether a hadith is authentic online and, for the weight of getting this wrong on purpose, the punishment for fabricating hadith.

Refs: Ṣaḥīḥ Muslim — site reader nos. 7374, 7376, 7377 (Imam Muslim's introduction).

🧭 The three citation patterns

1. Ṣaḥīḥ al-Bukhārī, followed by a number — the standard international numbering of al-Bukhārī. You can look the number up in the site's hadith reader or in a printed edition and find the same narration. The example used at the top of this page is Ṣaḥīḥ al-Bukhārī 2697, the narration on which the whole discipline of rejecting innovations rests, and it reads in full as below.
2. Ṣaḥīḥ Muslim — site reader no. … — an honest label for a technical fact. The site's Muslim corpus is numbered sequentially in reader order across 7,458 entries, Imam Muslim's introduction included, and that sequence does not match the standard international numbering of Ṣaḥīḥ Muslim. The number will take you to the right place in this site's reader; it should not be quoted elsewhere as though it were the printed edition's number. Where both Ṣaḥīḥs carry a narration, this site cites al-Bukhārī by preference for exactly this reason.
3. A Sunan collection, a number, and a stated grade — for Abū Dāwūd, at-Tirmidhī, an-Nasāʾī and Ibn Mājah, each narration in the repository carries its grade, written as graded ṣaḥīḥ in the site's hadith repository. Gradings are attributed to the repository rather than to a named scholar, and only ṣaḥīḥ narrations are used as evidence on rulings pages.

That narration is the reason a site like this one has to care about grading at all. If what is not from the religion is rejected, then a saying wrongly attributed to the Prophet ﷺ and acted upon is not a harmless extra — it is precisely the thing the hadith turns away. See Bidʿah, and the site's own sourcing standards at Editorial & Sourcing.