In the process of Power transformer quality control in whole life cycle, usually because the key is unknown, it will lead to unnecessary waste of resources. On the basis of statistical analysis, this paper divides the power transformer into seven parts and build a comparatively perfect risk level evaluation system, through the introduction of entropy and extension matter-element evaluation method, we integrate the risk level assessment with quantitative analysis to reduce the influence of subjective factors. It is a reasonable solution to the uncertainty and fuzziness of the evaluation process, At the same time it improves the traditional extension theory defects of excessive use of membership and rely on the maximum membership degree principle which lead to the result inaccurate. The component risk sorting has a certain guiding significance for the whole life cycle power transformer quality control .