It is common to ask what something means, or what someone means by something, but rare to ask what it means for something to mean something. What does ‘meaning’ mean? This is arguably the most fundamental question a human enquirer can ask, and it is quite difficult to answer. The prevailing view in the philosophy of language is that a sentence's meaning is defined by its truth-conditions, or the conditions under which it would be true. This truth-conditional semantic theory is simple yet explanatorily powerful; by starting with basic units of meaning and positing a simple set of rules and axioms, the theory can be used to derive an infinite amount of sentence meanings. Nevertheless, truth-conditional semantic theory also faces serious challenges; it fails to account for the meaning of non-truth-conditional utterances such as commands and questions, and it must rely on pragmatic processes to explain how sentence meaning is enriched to produce utterance meaning. In this paper, I put forth an alternative theory called satisfaction-conditional semantic theory, which accounts for the meaning of any utterance type, and which relies solely on pragmatic processes to arrive at utterance meaning, thereby avoiding the need to show how a truth-conditional sentence meaning is enriched to produce utterance meaning. This work has wide-reaching implications in cognitive science, artificial intelligence, and even law. By exploring how we use language to communicate meaning, we gain insight into the structure of the human mind, and we come to understand how words are more than just puffs of air; rather, they are pieces of cognitive machinery that constantly intercept and interact forcefully with events in the physical world.