<p>Lexical utilities have emerged as a promising way to model the precautionary principle in recent years. But some object that the lexical precautionary principle is hypersensitive because slight increases in risk of catastrophe can prompt it to recommend precautions regardless of cost. This article defends the lexical precautionary principle from the hypersensitivity objection by explaining why costs matter for what it recommends. In addition, we show how minimizing the probability of catastrophe tends to make the lexical precautionary principle insensitive to the precise dividing line between catastrophe and non-catastrophe. Finally, our discussion leads to a novel interpretation and justification of a <i>de minimis</i> condition for invoking the precautionary principle.</p>