Majority cycling and related social choice paradoxes are often thought to threaten the meaningfulness of democracy. Deliberation can protect against majority cycles—not by inducing unanimity, which is unrealistic, but by bringing preferences closer to single-peakedness. We present the first empirical test of this hypothesis, using data from Deliberative Polls. Comparing preferences before and after deliberation, we find increases in proximity to single-peakedness. The increases are greater for lower- versus higher-salience issues and for individuals who seem to have deliberated more versus less effectively. They are not merely a by-product of increased substantive agreement (which in fact does not generally increase). Our results are important, quite apart from their implications for majority cycling, because single-peakedness can be naturally interpreted in terms of an underlying issue dimension, which can both clarify the debate and allow a majority-winning alternative to be interpreted as a median choice and thus as an attractive “compromise.”