Emission trading is considered to be a cost-effective environmental economic instrument for pollution control. However, the policy design of an emission trading program has a decisive impact on its performance. Allowance allocation is one of the most important policy design issues in emission trading, not only for equity but also for policy performance. In this research, an artificial market for sulfur dioxide (SO2) emission trading was constructed by applying an agent-based model. The performance of the Jiangsu SO2 emission trading market was examined under different allowance allocation methods and transaction costs. The results showed that the market efficiency of emission trading would be affected by the allocation methods when the transaction costs are positive. The auction allowance allocation method was more efficient and had the lowest total emission control costs than the other three allocation methods examined. However, the use of this method will require that power plants pay for all of their allowance, and doing so will increase the production costs of power plants. On the other hand, output-based allowance allocation is the second best method. (C) 2012 Elsevier Ltd. All rights reserved.