We investigated the effect of different diffusion MRI acquisition protocols (incl. varying gradient-strength and signal-to-noise ratio) on the accuracy of axon diameter index (ADI) estimation in fiber space for human specimens. We used the Bland-Altman plot to estimate the error in the MRI-based ADI measurement. We found that gradient strengths above 900mT/m and b-values larger than b=50k s/mm2 provided accurate estimation of ADI (error smaller than 0.13 µm). The error for the best protocol was about 0.04 µm, allowing to distinguish white matter tracts with varying effective axon diameters (e.g. corpus callosum, genu, 1.5 µm and posterior body, 1.93 µm).