This paper presents a random forest (RF) classifier-based digital protection scheme which provides an effective discrimination between internal and external faults on a busbar. The measured current signals of all the bays (lines) connected to a busbar have been used as feature vectors. The system and fault parameters have been varied to generate a wide variety of simulation cases (33,600) consisting of both internal and external faults. By giving post-fault data of one cycle duration of all the bay currents as an input to the RF classifier, and taking only 30% of the total data (33,600) for training and remaining 70% of the total data for testing, an accuracy higher than 98% has been obtained. The PSCAD/EMTDC software package is used to model a prevailing 400-kV Indian busbar system for the purpose of authentication of the presented technique. The presented technique successfully differentiates between internal faults and external faults and remains unaffected against the change in system and fault parameters. In addition, the proposed scheme maintains stability under the Current Transformer saturation phenomena particularly during a heavy-through fault. A comparative analysis of the proposed scheme with the recently proposed scheme using support vector machine classifier clearly shows its superiority.