The horizontal curvature of railway tracks plays a significant role in vehicle‒track dynamics, and accurate knowledge of the track curvature is essential for vehicle operators to fully understand the running behavior of vehicles. However, it is generally difficult for wagon operators or researchers to obtain track geometry information due to confidential reasons. Using sensors installed on wagons is an alternative to estimate the track curvature. This paper proposes and compares two methods for estimating the track curvature: the gyro method and the GPS method. The gyro method calculates the curvature by analyzing the car body's yaw rate relative to its running velocity, while the GPS method utilizes the two-dimensional (2D) trajectory of the vehicle. First, a multibody system (MBS) simulation including a three-wagon train passing through a curve with a radius of 300 m is conducted, and it aims to study the influence of measured positions on the car-body and longitudinal compressive forces (LCFs) on the gyro method. A theoretical analysis of the GPS method highlights the significance of sampling distance and signal precision on the accuracy of curvature estimation. Secondly, an experimental study by installing a gyroscope sensor, a GPS sensor, and several accelerometers on a wagon is performed to compare the effectiveness of these two methods. The results that both methods can be used to estimate the track curvature and the gyro method has an obvious advantage when estimating the running curvature in complex track sections with many crossovers and parallel lines. Additionally, two applications are demonstrated: one involves utilizing the gyroscope and GPS sensors to determine the real running trajectory of the train, while the other integrates accelerometers to monitor the wheel‒rail interaction during a long journey.