To solve the problems of traditional risk assessment methods such as non-differential analysis of evaluation indexes

impossibility of risk evaluation with equal risk priority numbers（RPN）

and low accuracy of evaluation results

according to randomness and fuzziness of cloud model

we established a cloud model based on the failure mode and effect analysis（FMEA） of power transformer

and applied it to the practical risk assessment for power transformer.The assessment results show that the proposed method not only can effectively sort risk degree of main components of transformer

but also solve the impossibility of risk evaluation with equal RPN（such as core and OLTC）. So this proposed method can provide a convincing foundation for efficient maintenance strategy and the improvement of security and economy.