The improving quality of ultra-sensitive superconducting quantum interference devices has led to the construction of advanced full-tensor magnetic gradiometers (FTMGs) integrated on mobile carriers such as aircraft, vehicles and ships. FTMGs measure the smallest spatial magnetic gradients, possessing many practical advantages compared to conventional scalar surveys and three-component surveys. However, the magnetic interference field caused by the magnetic sources associated with the carrier significantly decreases the precision of measurement. In this paper, a novel method is proposed to compensate for the magnetic interference field. Specifically, a mathematical model containing compensation coefficients of the mobile FTMG system is constructed by extending the Tolles–Lawson model. By formulating the task into nonlinear problems, unknown compensation coefficients in the model can be solved by the flower pollination algorithm, which has the advantages of insensitivity to iterative initial values, less parameter tuning and strong global search abilities. Both the finite-element simulation results based on COMSOL Multiphysics and the results of a field experiment show that the proposed method provides a useful way for the optimal estimation of compensation coefficients and the reduction of the magnetic interference field.