The invention provides a super-short-term forecasting method for a load of a terminal of a power distribution network based on real-time measurement of a feeder end. The method comprises the following steps of obtaining real-time load data and history actual load data of the feeder end; carrying out super-short-term load forecasting on the load of the feeder end to obtain a super-short-term load forecasting value of the feeder end; obtaining real-time power data and a power forecasting value of a distributed power supply of the power distribution network and carrying out load summation calculation to obtain a to-be-distributed value of load forecasting; obtaining history load data of the power distribution network, analyzing the association relationship between the different types of loads and the load of the feeder end, calling a forecasting method base and obtaining forecasting results of different load types according to load classification; and carrying out weighted summation on the load forecasting results in a power distribution network area and removing redundancies to obtain a super-short-term load forecasting value for each power distribution network corresponding to the feeder end. According to the method, a super-short-term load forecasting model based on the load classification and the real-time measurement of the feeder end is constructed, so that the super-short-term load forecasting precision of the power distribution network is effectively promoted.