We review the experimental literature on collusion, focusing in particular on the roles of information and communication. We confront the results with the theoretical literature and discuss the policy implications. The main insights from the experi-mental literature are the following. In an environment in which firms are unable to communicate, experimental firms have little success in achieving collusion. The only mildly collusive results are achieved when only two firms are active. Increasing the amount of available information in general has little effect on the likelihood of collusion. The ability to communicate among sellers has a strong and positive effect on the ability to collude. This effect may be mitigated when buyers can communic...