Abstract: This paper presents a method for the predictive maintenance of distribution transformers. That is, a method of predicting which transformers are most likely to soon fail. Once predicted, such transformers may be subject to maintenance or replacement. This practice reduces the costs and increases the reliability of power distribution systems. This practice is common in transmission systems. In that domain, physical methods such as dissolved gas analysis see fantastic results. But such methods are cost prohibitive for distribution systems. Instead, this paper proposes to utilize a data driven framework for the task. Further, this framework only uses data which is readily available. Such data includes the transformer's specification, power loading, and weather-related information. Such data inspires the use of two suitable machine learning algorithms. The first is Random forests. The second is the Random Undersampling with AdaBoost (RUSBoost) algorithm. We test these algorithms on over 700,000 distribution transformers in Southern California. This test finds that both algorithms outperform the current state of practice. Further, it finds that the RUSBoost algorithm performs better than the Random Forest.