This paper generalizes the standard mean–variance paradigm to a mean–variance–ambiguity paradigm by relaxing the assumption that probabilities are known and instead assuming that probabilities are themselves random. It extends the CAPM from risk to uncertainty by incorporating ambiguity. This model makes the distinction between systematic ambiguity and idiosyncratic ambiguity and proves that the ambiguity premium is proportional to systematic ambiguity. It introduces a new measure of uncertainty that combines risk and ambiguity. Use of this model can be extended to other applications including portfolio selection and performance measurement.