A total of 4302 healthy blood donors were screened for elevated serum ferritin and transferrin saturation. Fifteen had increased serum ferritin at a follow-up examination. Five relatives of these donors also entered the study. Eleven patients had elevated liver iron concentrations, while five had normal liver iron concentrations. The R2 relaxation rate in the liver was first measured with a conventional multi-spin-echo imaging sequence, and then by a volume-selective spectroscopic multi-spin-echo sequence, in order to achieve a minimum echo time of 4 msec. No correlation was found between the relaxation rate R2 and the liver iron concentration, when R2 was calculated from the imaging data. Multi-exponential transverse relaxation could be resolved when the spectroscopic sequence was used. A strong correlation between the initial slope of the relaxation curve and the liver iron concentration was found (r = 0.90, p < 0.001). Signal intensity ratios between liver and muscle were calculated from the first three echoes in the multi-echo imaging sequence, and from a gradient echo sequence. A strong correlation between the logarithm of the signal intensity ratios and the liver iron concentration was found. Although both spectroscopic T2 relaxation time measurements and signal intensity ratios could be used to quantify liver iron concentration, the gradient echo imaging seemed to be the best choice. Gradient echo imaging could be performed during a single breath hold, so motion artifacts could be avoided. The accuracy of liver iron concentration estimates from signal intensity ratios in the gradient echo images was about 35%.