To better evaluate the prestress loss of in-service prestressed concrete bridge structures, this article takes the 32m prestressed concrete box girder, which is widely used in domestic high-speed railways, as the research object. 26 test model beams using C50 concrete were produced by converting the box girder into a T-shaped section at a scale of 1:16. The strain and elongation of the prestressed steel strand in the experimental beam were continuously and continuously monitored to determine the actual value of prestress loss, which was then compared and analyzed with the values calculated by commonly used prestress loss calculation specifications at home and abroad. The findings demonstrate that prestress loss mainly occurs in the early stage, which is instantaneous loss, accounting for about 70% of the total loss. 84.6% of the anchoring losses of the experimental beam account for over 70% of the total instantaneous losses, with a maximum of 80.57%. The long-term loss of the experimental beam fluctuates dramatically in the first 10 days after anchoring and gradually stabilizes in the later stage. Comparing the calculation results of the five specifications with the measured values in the experiment, the data shows that the calculation results of (TB 10002-2017) are the most consistent with the measured data in this experiment for long-term losses. The research results can provide reference for the calculation of prestress loss of prestressed beams in practical engineering.