The invention provides an electric charge recovery risk estimation method based on a user portrait, belongs to the power field, and solves a problem that the estimation result of a conventional methodis lower in accuracy. The method comprises the steps: extracting original data in a power utilization information collection system, a marketing system and a meteorological system, classifying the original data after the fusion and cleaning, and obtaining sample data; analyzing the sample data to form a power user label system; screening or calculating the data for electric charge recovery risk estimation from the sample data and the power user label system, and forming an electric charge recovery risk estimation label; carrying out the portraying of a power user according to the electric charge recovery risk estimation label, and forming an electric charge recovery risk portrait of the power user; training a C4.5 decision tree model according to the electric charge recovery risk estimation label and the electric charge recovery risk portrait, obtaining a C4.5 decision tree model for the electric charge recovery risk estimation, and carrying out the electric charge recovery risk estimation of the power user through the C4.5 decision tree model.