Students of public opinion have increasingly recognized issue framing as an important influence on political attitudes, but the precise means by which frames affect attitudes is not well understood. We argue that one distinctive way in which frames affect attitudes is by influencing the importance individuals attach to issue-relevant beliefs. We contrast this mechanism with the more familiar means of persuasion via change in belief content. Data come from two laboratory experiments. In each, a controversial issue was framed in one of two ways. We measured framing's influence on belief content, belief importance, and issue opinion. In both experiments, framing significantly affected issue opinion. Causal analysis shows that framing independently affected belief content and belief importance, and that each contributed to issue opinion.