In this paper, we propose a representation of excuses in the context of multiagent systems. We distinguish five classes of excuses, taking as starting point both jurisprudential and philosophical studies about this topic, and we discuss their acceptance criteria. We highlight the following classes of excuses: epistemic excuses, power-based excuses, norm-based excuses, counts as-based excuses and social-based excuses and we represent them using social dependence networks. The acceptance criteria individuate those excuses which success in maintaining the trust of the other agents, e.g. in the context of social networks, excuses based on norms seem better than counts as-based ones in achieving this aim.