The invention provides a big data-based three-phase unbalance analysis and load adjustment method of a distribution transformer. The method comprises the steps of (100) classifying distribution transformer districts; (200) extracting data from a power system; (300) carrying out correlation analysis on a power fluctuation; (400) positioning the distribution transformer districts with strong correlation between a three-phase unbalance fluctuation and a low-voltage distribution transformer district line loss fluctuation and looking up a low-voltage user with strong power fluctuation correlation; (500) verifying and judging the reason of a three-phase unbalance and carrying out three-phase load adjustment; and (600) evaluating a loss reduction effect, turning to the step (200) if the loss reduction effect is invalid, and completing the steps if the loss reduction effect is valid. The big data-based three-phase unbalance analysis and load adjustment method of the distribution transformer can meet the requirements of enterprise power production, operating management and power service in the aspect of improving the quality, the benefits and the efficiency, and optimal configuration and efficient service of power resources are promoted.