Can information provision reduce the risks associated with irregular migration? We address this question conducting a large-scale experiment with about 7,000 secondary school students in Guinea. Combining aggregate statistics and video-testimonies by migrants who settled in Europe, we study the effect of three information treatments: (i) about risks and costs of the journey; (ii) about economic outcomes in the destination country; and (iii) a treatment pooling (i) and (ii). We find that one month after the intervention, all three treatments affect beliefs about the risks and the economic outcomes of migration. However, one year after the intervention, only the first has a significant effect on migration outcomes: providing information about the risks and costs of the journey reduces international migration by 51%. The effect is driven by a decrease in migration without a visa (i.e., potentially risky and irregular). Furthermore, the reduction is larger for students who, at baseline, underestimated the risks connected to international migration.