When a large-scale disaster occurs, a set of relief centers should be determined to accommodate evacuees and a variety of multi-commodity should be distributed to these relief centers to provide basic life support. Because the multi-commodity distribution at peacetime may be imperfect and unbalanced, the surplus commodities in some relief centers can be redistributed to other relief centers with shortages, to make the effective and efficient use of these commodities. This multi-commodity redistribution problem is also an important issue in the emergency management. Various uncertain elements include transportation network, supply and demand, making this problem a big challenge. To handle this problem, a two-stage mixed-integer stochastic programming model was proposed to facilitate this multi-commodity redistribution process. In our model, we define the dissatisfaction cost based on the relief center size, unmet demand and oversupply of commodity in the relief center. Then, our objectives are to minimize the total dissatisfaction cost in the first stage and minimize the total transportation time in the second stage, sequentially. Finally, a randomly generated numerical instance is tested and computational results show that the proposed model can provide effective and efficient decisions in the multi-commodity redistribution process.