This paper investigates how attitudes toward the United States are affected by the provision of information. We generate a panel of attitudes in urban Pakistan, in which respondents are randomly exposed to fact-based statements describing the United States in either a positive or negative light. Anti-American sentiment is high and heterogenous in our sample at the baseline, and systematically correlated with intended behavior, such as intended migration. We find that revised attitudes are, on average, significantly different from baseline attitudes: attitudes are revised upward (downward) upon receipt of positive (negative) information, indicating that providing information had a meaningful effect on U.S. favorability. There is, however, substantial heterogeneity in the revision of attitudes, with a substantial proportion of individuals not responding to the information. Nonrevisions are primarily a result of nonmalleability of attitudes. Revisions are driven by both saliency bias and information-based updating. In addition, the information-based updating is partly consistent with unbiased belief updating.