This work examined the accuracy and reproducibility of ultra-short TE (UTE) R2* mapping in patients with liver iron overload. Fifteen subjects with known or suspected liver iron overload were scanned at 1.5T and 3.0T using a radial UTE, two Cartesian multi-echo, gradient-echo acquisitions, and an R2-based (FerriScan) reference acquisition. UTE R2* measurements demonstrated excellent reproducibility across field strengths (with expected linear increase with field strength) and high correlation with liver iron concentration. Cartesian approaches offered excellent reproducibility for R2*<1000s-1. However R2*>1000s-1, neither Cartesian approach were reproducible across field strength, suggesting that the range of R2* had been surpassed.