PURPOSE: To present a novel technique for high-resolution stimulated echo diffusion tensor imaging with self-navigated interleaved spirals readout trajectories that can inherently and dynamically correct for image artifacts due to spatial and temporal variations in the static magnetic field (B0) resulting from eddy currents, tissue susceptibilities, subject/physiological motion, and hardware instabilities. METHODS: The Hahn spin echo formed by the first two 90° radiofrequency pulses is balanced to consecutively acquire two additional images with different echo times and generate an inherent field map, while the diffusion-prepared stimulated echo signal remains unaffected. For every diffusion-encoding direction, an intrinsically registered field map is estimated dynamically and used to effectively and inherently correct for off-resonance artifacts in the reconstruction of the corresponding diffusion-weighted image. RESULTS: After correction with the dynamically acquired field maps, local blurring artifacts are specifically removed from individual stimulated echo diffusion-weighted images and the estimated diffusion tensors have significantly improved spatial accuracy and larger fractional anisotropy. CONCLUSION: Combined with the self-navigated interleaved spirals acquisition scheme, our new method provides an integrated high-resolution short-echo time diffusion tensor imaging solution with inherent and dynamic correction for both motion-induced phase errors and off-resonance effects.