We address the problem of intention and capability calibration in human-robot collaboration with a decision-theoretic approach.