This article argues that a fragment from a lost treatise by Abū Bakr al-Rāzī (d. 925) is preserved in the Book on Morphology Kitāb al-Taṣrīf) by Ps-Ǧābir ibn Ḥayyān. Paul Kraus reached the conclusion that the collection to which this book belongs was written between the end of the ninth and the beginning of the tenth century AD. This fragment represents the first attempt – to our knowledge – to analyze the logical structure of sign-based inference in Arabic, which is (...) known as istidlāl bi-al-šāhid ʿalā al-ġāʾib among theologians and philosophers. The author distinguishes between sign-inferences based on homogeneity (al-muǧānasa), course of habit (maǧrā al-ʿāda) and traces (āṯār). After providing a translation of the fragment, the first part of this paper argues that its author is Abū Bakr al-Rāzī. My argument is based on a comparison between this text and a passage from the Doubts About Galen, which is also by Abū Bakr al-Rāzī. I hypothesize that at least two other fragments from the same work or from different works by Abū Bakr al-Rāzī are preserved in the corpus attributed to Ps-Ǧābir. The second part of the paper aims to reconstruct Abū Bakr al-Rāzī’s theory of sign-inference. In so doing, I show the historical influence that Hellenistic debates on sign-inference might have had on al-Rāzī, and I situate al-Rāzī’s theory in the context of the prominent use that the theologians of the kalām made of the istidlāl bi-al-šāhid ʿalā al-ġāʾib. To offer a more comprehensive reconstruction of al-Rāzī’s theory of sign-inference, this article compares the critical approach presented in the newly identified fragment with the epistemological framework outlined in the Doubts About Galen. Finally, this article shows that Abū Bakr al-Rāzī’s theory of sign-inference had a strong influence on al-Fārābī’s logical developments especially in his Epitome of the Prior Analytics, even if he does not acknowledge this intellectual debt. (shrink)