To build an effective assessment method for the risk of electric power transformers, the full condition model of transformer based on condition-based maintenance (CBM) and Markov Process is established, considering the complexity and uncertainty of transformer fault. The research method of classified index is used for confirming the severity of transformer defects. Combined with the calculation model for subcomponent fault frequency, entropy-weighted fuzzy method is adopted to quantify the serious degree. Based on the above research the assessment matrix is established. Then a new assessment method for the risk of electric power transformers is given. The simulation program is compiled, and a case is analyzed by the risk assessment technology. The calculation result concludes the risk value and maintenance strategy. Empirical results show that the approach proposed is available and effective.