In this study, in-service crossbeam of steel bridge piers, which are designed to be reinforced or dismantled, were measured for internal stress using the magnetostriction method in the field. Stress analysis was conducted using the shear stress difference integration method to determine the bending and shear stress distributions occurring in 3 sections of a crossbeam. Furthermore, the variation in the measured stress distribution and the technical issues of magnetostriction methods arising from on-site measurements are discussed. The measuring object in this study is a continuous crossbeam with rectangular box section supported by 3 piers. Although it is desirable to use a calibration curve based on the same steel material as that being measured, in this study, a calibration curve was created by conducting tensile tests on a new steel material of the same type. On-site measurements were carried out at three cross-sections of the crossbeam, near the center spans and an intermediate support. The measurement and analysis results indicate that the current analysis results exhibit variations at several levels of around 10 MPa. Additionally, there is a tendency for the analysis values to converge toward the estimated values as one progresses in the direction of integration. I believe this is caused by the shear stress integral method assuming a vertical stress level of 0 MPa (free end) at the endpoints. Therefore, in the case of a box girder section, which consists of plates without free ends, I suggest applying the assumed values at the endpoints in the analysis to the actual values measured using the one-point measurement mode.