Using data from a two-year randomized experiment in rural China, this paper studies the influence of social networks on the decision to adopt a new weather insurance product and the mechanisms through which social networks operate. In the first year, I provided financial education to a random subset of farmers and found a large social network effect on insurance take-up: for untreated farmers, having an additional friend receiving financial education raises take-up by almost half as much as obtaining financial education directly, a spillover effect equivalent to offering a 12% reduction in the average insurance premium. By varying the information available to subjects about their peers’ take-up decisions and using randomized default options, I show that the positive social network effect is not driven by scale effects, imitation, or informal risk-sharing, but instead by the diffusion of insurance knowledge. One year later, social networks continue to affect insurance demand: observing an above-median share of friends receiving payouts increases insurance take-up at a rate equivalent to about 50% of the impact of receiving payouts directly. I also find that social network effects are larger in villages where households are more strongly connected, and when the people who receive financial education first are more central in the social network.