During the past many years, Lanchester’s differential equations have been the norm for computer simulations of combat. However, these methods cannot fully reflect the inherent uncertainty of related variables in modeling. In view of this limitation, we propose a Lanchester’s combat models based on cloud qualitative simulation. This method introduces numeric characteristics and cloud measurement index of the normal cloud to represent the qualitative landmarks, and propagate the cloud measurement index within constraint sets to improve the accuracy of simulation results. Consequently, this method allows for a more accurately representation and interpretation of the uncertainty within related variables in modeling, and expands the scope of warfare application. Especially when the relevant variables of the model cannot or need not be quantified precisely involving with the case of representation of qualitative variables with uncertainty, this method has certain advantages. Finally, the examples of classic historical battles demonstrate that this method is a correct and feasible method.