The primary goal of MR in acute stroke imaging is to determine the ischemic tissue at risk. Since a narrow time window is available for treatment, an accessible point-of-care system is desirable. Ultra-low field MR (ULF) shows promise but suffers from low signal-to-noise ratios (SNR). We performed a theoretical study of the SNR and contrast-to-noise (CNR) ratios for a range of B0 fields, with corresponding tissue relaxation times, considering different gradient performances. Illustrative simulated DWI brain images are provided. Results suggest that despite the challenges, ULF is promising for imaging acute stroke lesions.