This article examines Śāntarakṣita’s and Kamalaśīla’s understandings of svatantra and prasaṅga proofs in the attempt to clarify how and why Tibetan Prāsaṅgikas came to portray svatantra inference as an instance of the very thing Madhyamaka rejects. The article proceeds in four parts. A brief comparison of Patsap Nyimadrak’s portrayal of svatantra inference with Bhāviveka’s and Candrakīrti’s employment of this expression shows that Patsap expanded the meaning of it, charging its users with embracing a realism at odds with Madhyamaka emptiness. Patsap’s (...) arguments against Śāntarakṣita’s “neither one nor many” proof lead us consider the latter’s and Kamalaśīla’s understanding of svatantra and prasaṅga proofs, which turns out to be consistent with, if more sophisticated than, Bhāviveka’s and Candrakīrti’s usage. An exploration of Kamalaśīla’s development of the proof of the naturelessness of all things, real and imagined, calls into question the svatantra-prasaṅga distinction but still leaves us without a source for Patsap’s characterization. A consideration of Śāntarakṣita’s and Kamalaśīla’s use of svatantra to characterize Mīmāṃsā conceptions of an authorless Veda and “intrinsic validity,” as well as their own view of “self-awareness,” offers potential sources for characterizing valid cognition and, by extension, inference as “autonomous.”. (shrink)