The concept of ‘‘certitude” (yaqīn) is central in Arabic discussions of the theory of demonstration advanced by Aristotle in the Posterior Analytics. In the Arabic tradition it is ‘‘certitude,” rather than ‘‘knowledge” (‘ilm=Greek epistēmē), that is usually identified as the end sought by demonstrations. Al-Fārābī himself devotes a short treatise, known as the Conditions of Certitude (šarā’iṭ al-yaqīn), to determining the criteria according to which a subject can claim to have absolute certitude of any proposition. In this article the author traces the roots of the concept of certitude to the Arabic translation of the Posterior Analytics. She then offers a close reading of the text of al-Fārābī's Conditions and its parallels in his Epitome of the Posterior Analytics (Kitāb al-Burhān). She argues that the identification of certitude as a cognitive state distinct from knowledge enabled al-Fārābī to introduce a number of interesting and sometimes problematic innovations into Arabic epistemology. Foremost amongst his positive innovations is the recognition that contingent as well as necessary propositions allow for an attenuated form of certitude, and the introduction of reliabilism into the justification of claims to certitude. But al-Fārābī is also one of the first philosophers to articulate explicitly the idea that certitude must include an element of ‘‘knowing that one knows,” a controversial principle that tends to undermine both al-Fārābī’s reliabilism and his efforts to loosen some of the strictures in traditional Aristotelian epistemology.