This paper proposes a Real Time Load Estimator (RTLE) for the obtaining of accurate information on the realtime operating state of Distribution Systems (DSs) feeders. The RTLE is based on off-line information (customers consumption and typical load curves); a computationally efficient load-flow algorithm that uses a data structure named Node-Depth Encoding, and available real-time measurements. Using only offline information, the proposed RTLE enables the off-line estimation of the loads of distribution transformers (aggregated load curves) and provides a measure of uncertainty of the estimation. It also refines the off-line estimates by processing the available real-time measurements via a computationally efficient load flow algorithm. Aspects of interest as impact of type and accuracy of measurements and the presence of measurements with gross errors are analyzed. The simulation results show the efficiency of the proposed RTLE in a test feeder.