There are many situations where the best outcome is reached through co-operation and co-ordination of agents’ actions. Although this is the best collective outcome, economic agents may fail to implement such co-operative strategy. The reason for this failure may be lack of information about the gains from co-operation, or lack of capacity to implement the co-operative strategy. The present work focuses on two obstacles to co-operation that are linked with the incentives of the economic agents, and that are present even when the problems of information and capacity are taken care off. The two obstacles are the incentive to free ride and the strategic risk. The former stems from the possibility of obtaining gains without paying the associated costs (which are incurred by the agents that decide to co-operate); the latter is the risk of being the only one (or among the few) that acts co-operatively, so that the agent pays the costs but obtains less than what it would be feasible had other agents decided to co-operate. In this setting, using a game theoretical approach, we distinguish several cases of co-operation failures according to the relevance of those two obstacles. The analysis is then applied to contractual design and financial incentives. The overall message is the importance of identifying the source of co-operation failure in order to devise an effective policy to induce co-operation. It may not be enough to tell people (and institutions) that they should co-operate because it is in their interest, it is necessary to identify the incentives that shape agents’ decisions and are responsible for co-operation failures.