Despite significant efforts, obesity continues to be a major public health problem, and there are surprisingly few effective strategies for its prevention and treatment. We now realize that healthy diet and activity patterns are difficult to maintain in the current physical environment. Recently, it was suggested that the social environment also contributes to obesity. Therefore, using network-based interaction models, we simulate how obesity spreads along social networks and predict the effectiveness of large-scale weight management interventions. For a wide variety of conditions and networks, we show that individuals with similar BMIs will cluster together into groups, and if left unchecked, current social forces will drive these groups toward increasing obesity. Our simulations show that many traditional weight management interventions fail because they target overweight and obese individuals without consideration of their surrounding cluster and wider social network. The popular strategy for dieting with friends is shown to be an ineffective long-term weight loss strategy, whereas dieting with friends of friends can be somewhat more effective by forcing a shift in cluster boundaries. Fortunately, our simulations also show that interventions targeting well-connected and/or normal weight individuals at the edges of a cluster may quickly halt the spread of obesity. Furthermore, by changing social forces and altering the behavior of a small but random assortment of both obese and normal weight individuals, highly effective network-driven strategies can reverse current trends and return large segments of the population to a healthier weight.