This paper presents a magnetic flux density estimation method based on a random forest (RF) in the vicinity of high-voltage overhead lines. RF is a group of single decision trees where each is created with a different, randomly chosen subset of the training data and with a randomly chosen subset of the features at every node. Each node in the tree represents a feature from the input dataset, each branch a decision, and each leaf at the end of a branch the corresponding output value. The developed method was evaluated by comparing its performance with the testing dataset and validated through multiple approaches. Validation was conducted by comparing the results obtained using the proposed method with results obtained by the method based on Biot-Savart (BS) law and also with the measurement results. A graphical comparison of the calculated and measured values is presented for a case of a real 400 kV high voltage transmission line with a horizontal phase conductor configuration. Furthermore, validation was performed by comparing the results of the RF method with measurement results which were conducted on 22 different high-voltage transmission lines. The results demonstrate that the proposed method based on the RF provides results comparable to other standard methods, confirming its reliability for determining the spatial distribution of magnetic flux density.