Magnetic Particle Imaging (MPI)-guided Magnetic Fluid Hyperthermia (MFH) has the potential for widespread utilization, as it allows for the prediction of magnetothermal dosage, real-time visualization of the thermal therapy process, and precise localization of the lesion area. However, the existing MPI-guided MFH (MPI-MFH) method is insensitive to concentration gradients of magnetic nanoparticles (MNPs) and is susceptible to causing damage to normal tissues with high MNP concentrations during MFH treatment, while inadequately heating tumor tissues with lower MNP concentrations. In this work, we established a relationship between MNP concentration and heating efficiency through simulations and phantom measurements, enabling the optimal selection of MFH parameters guided by MPI. Based on these findings, we developed a high-gradient field MPI-MFH method using a field-free point (FFP) approach to achieve precise local heating. Phantom experiments and in vivo glioma model experiments were conducted to validate this proposed method. The results demonstrated that the proposed method of MPI-MFH can improve the MNP concentration gradient sensitivity to ±1 mg/ml, thereby enabling more effective lesion-site heating without damaging normal tissues. This method not only reduced glioma size effectively but also holds promise for application in various other types of cancers.