Ambiguity attitudes have been prominently used in economic models, but we still know little about their demographic correlates or their generalizability beyond the West. We analyze the ambiguity attitudes of almost 3,000 students across thirty countries. For gains, we find ambiguity aversion everywhere, while ambiguity aversion is much weaker for losses. Ambiguity attitudes change systematically with probabilities for both gains and losses. Much of the between-country variation can be explained through a few macroeconomic characteristics. In contrast, we find massive unexplained variation at the individual level. We also find much unexplained heterogeneity in individual responses to different decision tasks.