Scholars of Islamic law point to the absence of any extant work of legal theory between the Risāla of al-Shāfiʿī and the Fuṣūl of al-Jaṣṣāṣ as a major barrier to reconstructing the history of Islamic legal thought. However, careful analysis of three major works of the Ḥanafī jurist al-Ṭaḥāwī, Aḥkām al-Qurʾān, Sharḥ maʿānī al-āthār and Sharḥ mushkil al-āthār, reveals the existence of myriad brief passages elaborating questions of legal theory scattered throughout their many volumes. This study reconstructs the legal thought of al-Ṭaḥāwī as a window onto legal theory in the late 3rd/9th and early 4th/10th centuries, a crucial period of transformation between late formative and post-formative Islamic law. It argues that al-Ṭaḥāwī’s works are not direct precursors to the genre of uṣūl al-fiqh, but instead represent a different, previously unrecognized, type of intellectual and literary activity. This activity, here termed practical hermeneutics, is concerned with demonstrating in detail how individually coherent rules of law may be derived from the often messy texts of revelation. The integrated reading of al-Ṭaḥāwī’s entire hermeneutical corpus uncovers several areas in which his legal thought departs quite notably from that of other jurists, suggesting that al-Ṭaḥāwī was neither as dependent on al-Shāfiʿī nor as closely related to mature uṣūl al-fiqh as has been suggested in previous studies. Most crucially, al-Ṭaḥāwī’s works unsettle accepted accounts of Islamic legal theory which assign varying levels of authority to a series of clearly distinguished legal sources—Qurʾān, Sunna, consensus, etc. This study demonstrates that, in contrast to both al-Shāfiʿī and later uṣūlīs, al-Ṭaḥāwī’s legal thought blurs boundaries between these categories and instead rests upon an underlying binary concept of legal authority which draws a crucial distinction between knowledge that might permissibly be reached by inference, and knowledge that can only have come from revelation. The authority that al-Ṭaḥāwī grants any given source is therefore not a function of its formal characteristics, but rather the result of his own judgment about content and origins.