Failure mechanisms of power transformers are complex, and the determination of fuzzy membership function is subjective and fails to take randomness into account. Beside, dissolved gas analysis (DGA) alone does not provide sufficient information to predict potential failures of a transformer. Aiming at these problems, an assessing model of transformer insulation is proposed base upon fuzzy cloud theory. Based on the determination of variable index weights, insulation condition system was established and insulation grade was classified. The cloud model was used to describe the fuzziness and randomness of the transformer and assess the condition of transformer insulation combined with infrared thermal imaging method and DGA method. Case studies and comparison analyses with other assessing method show that fuzzy cloud model analysis method is feasible and effective, and the method also offers a new way of assessment for transformer insulation condition.