Low‐field (<1 tesla) MRI scanners allow more widespread diagnostic use for a range of cardiac, musculoskeletal, and neurological applications. However, the feasibility of performing robust fMRI at low field has yet to be fully demonstrated. To address this gap, we investigated task‐based fMRI using a highly sensitive transition‐band balanced steady‐state free precession approach and standard EPI on a 0.55 tesla scanner equipped with modern high‐performance gradient coils and a receive array.

TR and flip‐angle of transition‐band steady‐state free precession were optimized for 0.55 tesla by simulations. Static shimming was employed to compensate for concomitant field effects. Visual task‐based fMRI data were acquired from 8 healthy volunteers. For comparison, standard EPI data were also acquired with TE = . Retrospective image‐based correction for physiological effects (RETROICOR) was used to quantify physiological noise effects.

Activation was robustly detected using both methods in a 4‐min scan time. Transition‐band steady‐state free precession was found to be sensitive to interference from subtle spatial and temporal (field drift, respiration) variations in the magnetic field, counteracting potential advantages of the reduced magnetic susceptibility effects compared to its utilization at high field. These adverse effects could be partially remedied with static shimming and postprocessing approaches. Standard EPI proved more robust against the sources of interference.

BOLD contrast is sufficiently large at 0.55 tesla for robust detection of brain activation and may be employed to broaden the spectrum of applications of low‐field MRI. Standard EPI outperforms transition‐band steady‐state free precession in terms of signal stability.