<p>This paper develops an argument for classical theism from what I call recursive intelligibility, which is the open-ended, convergently unifying, and domain transcendent applicability of mathematics to the structure of physical reality. The argument first distinguishes local intelligibility, or the cognitive-reality alignment that evolutionary selection plausibly explains, from the global and recursive intelligibility at issue, namely the continuing success of abstract mathematical structures in domains maximally remote from evolutionary relevance. Cases such as the later physical applicability of Riemannian geometry, complex analysis, and Lie group theory exhibit a form of intelligibility that extends well beyond what the evolutionary byproduct story straightforwardly predicts. I argue that classical theism, specifically the logos tradition on which reality reflects a rational source, generates a genuine expectation of recursive intelligibility, whereas evolutionary naturalism predicts only local intelligibility and faces explanatory difficulties with domain-transcendence, convergent unification, and the apparent absence of an intelligibility horizon. I distinguish the position from fine-tuning arguments, Plantinga's evolutionary argument against naturalism, and from standard formulations of the argument from reason, and answer objections from anthropic reasoning, instrumentalist accounts of mathematics, the missing base rate of applicable mathematics, and Tegmark's mathematical universe hypothesis. My conclusion is that recursive intelligibility is more probable given classical theism than given evolutionary naturalism.</p>