PURPOSE: To develop a B 0 $$ {B}_0 $$ self-navigation approach to estimate respiratory motion for motion-corrected liver T 1 $$ {T}_1 $$ mapping using a Look-Locker acquisition with radial stack-of-stars trajectory. METHODS: The proposed method derives 1D field-map profiles from the oversampled k-space center to estimate a normalized breathing curve and the B 0 $$ {B}_0 $$ variation amplitude for each slice and coil. B 0 $$ {B}_0 $$ drift and contrast variations, inherent to the Look-Locker acquisition, were modeled and corrected by fitting and demodulating drift and offset terms. The breathing curve was employed to bin data into motion states for motion-resolved reconstruction, followed by water-specific T 1 $$ {T}_1 $$ mapping. Simulations with an anatomical body model and in vivo experiments with a Look-Locker multi-echo gradient echo sequence were performed to validate the technique. The estimated normalized breathing curve was compared with magnitude- and phase-based self-navigation approaches using principal component analysis. RESULTS: The proposed B 0 $$ {B}_0 $$ self-navigation reliably estimated the normalized breathing curve and the B 0 $$ {B}_0 $$ variation amplitude in simulations and in vivo. B 0 $$ {B}_0 $$ variation amplitudes increased with greater tissue displacement, with median values across slices and coils ranging from 4 to 15 Hz at 3 T in volunteers. Motion-resolved reconstruction using the estimated breathing curve reduced motion artifacts and improved image and T 1 $$ {T}_1 $$ mapping quality compared to motion-averaged reconstruction. CONCLUSION: B 0 $$ {B}_0 $$ self-navigation allows estimation of respiratory motion in acquisitions with varying contrast and quantifies the B 0 $$ {B}_0 $$ variation amplitude, providing a possible surrogate signal for tissue displacement and enabling self-gated liver T 1 $$ {T}_1 $$ mapping using a Look-Locker approach.