The truth bi-conditionals are the expressions ‘A iff T[A]’ where A is a sentence, T is a predicate (the ‘truth’ predicate) and [A] a name for A according to some formal representation of syntax (such as Gödel coding). Theories defined in terms of truth bi-conditionals are typically either inconsistent or deductively, and conceptually, simple. As observed already by Tarski, compositional truth axioms, such as “for all A, B: T[A and B] iff T[A] and T[B]” are not derivable from the basic bi-conditionals except in trivial cases. Nevertheless, Quine, Horwich and others have proposed that the truth bi-conditionals are all there is to truth. In this talk I present proof-theoretic support for this extreme view and show how remarkably strong systems are implicit in weak assumptions vis-à-vis truth.