The invention discloses a method for evaluating the running health state of a magnetic control paralleling reactor. The method comprises the steps that (1), the running characteristic of the magnetic control paralleling reactor is considered, indexes for evaluating the running health state of the magnetic control paralleling reactor are specifically selected, and the measured values of the evaluation indexes are obtained; (2), normalization processing is carried out on the measured values of the evaluation indexes through reactor preventive test regulations and the data of field researches, and evaluation index quantized values of the same measuring unit are obtained; (3), solution of the state grade membership degree is carried out on the evaluation index quantized values based on the fuzzy theory, and the different state grade membership degrees of all the evaluation indexes are obtained; (4), the running health state of the magnetic control reactor is comprehensively evaluated through an evidence theory evaluation model, and the running state of the magnetic control reactor is determined. According to the method, sample learning does not need to be carried out through a large number of prior data, and the running state of the reactor can be accurately diagnosed.