The phrase “saheeh hadeeth” may refer to one of two things:
In general terms it includes those which are mutawaatir (narrated from so many by so many in each stage of transmission that it is inconceivable that they could all have agreed upon a lie), saheeh li dhaatihi (sound in and of itself), saheeh li ghayrihi (sound because of corroborating evidence) and hasan (good).
Al-Haafiz ibn Hajar said:
Most of the scholars of hadeeth do not differentiate between hasan and saheeh. End quote from al-Nukat (1/480),
In specific terms it includes saheeh li dhaatihi (sound in and of itself) and saheeh li ghayrihi (sound because of corroborating evidence) only.
Based on this definition, a saheeh hadeeth is one which is narrated by men of good character, who are known for their good memories and precision, with a continuous isnaad, and is not odd or faulty.
If the precision is lacking and is not complete, then it is hasan li dhaatihi (hasan in and of itself). If it has a number of isnaads, then it is saheeh li ghayrihi (saheeh because of corroborating evidence).
See Nakhbat al-Fikr by al-Haafiz Ibn Hajar (may Allaah have mercy on him).
From this definition we may sum up the conditions of a saheeh hadeeth as follows:
1 – Good character of all its narrators
2 – Good memory and precision on the part of narrators with regard to what they are narrating.
3 – Continuous isnaad from beginning to end, meaning that each narrator heard the hadeeth from the one before him.
4 – The hadeeth is free from any oddness in its isnaad or text. What is meant by “odd” is anything in which the narrator narrates something that contradicts the narration of a sounder narrator.
5 – The hadeeth is free from faults in its isnaad and text. A “fault” is a subtle problem that undermines the soundness of the hadeeth, which can be detected only by the well versed scholars of hadeeth.
The definition of these conditions came about as the result of the later imams studying the words of the scholars of hadeeth and their applications. Hence you may find things in the words of the earlier scholars which point to these conditions.
For example: Imam al-Shaafa’i (may Allaah have mercy on him) said in al-Risaalah (370-371):
Evidence cannot be established on the basis of a report narrated by a few unless several factors are present, such as:
The one who narrated it is trustworthy in his religious commitment, known to be truthful in his speech, understanding what he narrates, and knowledgeable about the wording and possible interpretation of the hadeeth; and he should be one of those who can narrate the hadeeth exactly as he heard it, not based on the meaning but with the exact wording, because if he if narrated on the basis of meaning and not with the exact wording, and he does not have knowledge of possible interpretations, he may inadvertently change what is halaal into haraam. But if he narrates it exactly, there is no fear that it may be changed.
And he should know the hadeeth very well, if he is narrating from memory or he should take get care of his book if he is narrating from his book. If he checks what he knows with the scholars of hadeeth, he should be in agreement with them, and he should not be mudallis, i.e., one who narrates from one who met (a narrator) but did not hear it from him, or who narrates from the Prophet (peace and blessings of Allaah be upon him) something that contradicts the narration of authentic scholars from the Prophet (peace and blessings of Allaah be upon him). The same must be true for the narrators who came before him (in the isnaad), who narrated it to him, until the hadeeth ends with an uninterrupted chain all the way back to the Prophet (peace and blessings of Allaah be upon him) or to the one who narrated it from the Prophet (peace and blessings of Allaah be upon him). End quote.
If all these conditions are met then a hadeeth is saheeh according to scholarly consensus, as was narrated by Ibn al-Salaah (may Allaah have mercy on him).
See: al-Muqaddimah fi ‘Uloom il-Hadeeth (8) and al-Dhahabi in al-Mooqidah, (24).
Some of the scholars did not list all these conditions:
Imam Maalik and Abu Haneefah accepted mursal hadeeth, which is compromising the condition of having a continuous isnaad all the way to the source of the hadeeth.
Some scholars accepted mudallas hadeeth even if the narrators did not state that they heard it.
Al-Dhahabi said in al-Mooqidah (24): The scholars of hadeeth added a condition that the hadeeth should be free from any oddness or fault. This is subject to further discussion according to the principles of the fuqaha’, because many of what may be considered as faults are not accepted as such by them.
See: Tadreeb al-Raawi (1/68-75, 155).
What is meant is that the scholars’ differences with regard to classing ahaadeeth as saheeh are due to two reasons:
1 – Their differences regarding some of the conditions of soundness. That is because those who compromised on some of these conditions were bound to class as saheeh reports that others did not class as saheeh.
2 – Their differences in applying these conditions to a specific hadeeth. They may differ as to the good character of some narrators or whether the isnaad is uninterrupted and so on.
It should be noted that these conditions are based on evidence from sharee’ah and evidence based on reason. These conditions are not just a ritualistic procedure, rather they are rational and serve a clear purpose. They are no less than the result of the efforts of thousands of scholars and the result of the tremendous efforts of the earlier scholars of hadeeth during the long years when the books of hadeeth were being compiled in the first three centuries and those who came after them.
Anyone who wishes to know more may read the book by al-Khateeb al-Baghdaadi entitled al-Kifaayah fi ‘Ilm al-Riwaayah
And Allaah knows best.