For the risky multi-criteria decision-making problem in which the criteria value is linguistic variables, and the weighted coefficients are partially known, a decision-making approach based on the prospect theory and the cloud model is proposed. An improved method for generating cloud is proposed, and the linguistic variables are converted into the cloud model based on this method. After that, the cloud distance and the cloud comparison rule are defined. Then, the prospect value function of the cloud model can be defined, and the prospect decision matrix is constructed based on all other alternatives as the dynamic reference point. On the basis of the optimal criteria weights attained by the algorithm of maximizing deviation, the order of alternatives can be listed by comparing the integrated prospect values of each alternative. Finally, an example is illustrated to verify the rationality and reliability of the proposed method.