The invention discloses a multiple state variable prediction-based voltage transformer state assessment method comprising the following processes: main fault state variables of a voltage transformer are chosen; the chosen main fault state variables of the voltage transformer are classified into four types: static parameters, dynamic parameters, quasi-dynamic parameters and external parameters; based on the classified main fault state variables of the voltage transformer, a state variable prediction algorithm is adopted for predicting a state of the voltage transformer, and a predicted state value can be obtained; a neural network prediction algorithm is adopted for determining a trend of an operation state of the voltage transformer to be assessed, and a comprehensive assessment result of a voltage transformer state is given based on a real time operation state and a real time trend of the voltage transformer. The method disclosed in the invention is advantageous in that a large amount of state information of the voltage transformer can be subjected to comprehensive analysis, and potential faults of the voltage transformer can be discovered and predicted.