High purity iron samples were torsion strained and subsequently annealed. The 2-D microstructure images were quantitatively analysed. The results were used as an input to two stereological methods for spatial (3-D) grain size distribution reconstruction. The study showed that pure iron samples SPD processed and annealed for a short time could be successfully characterized by a log-normal distribution of 3-D grain size. The modeling method showed a significantly better fit in terms of chi2 values than Scheil-Schwartz-Saltykov method. Having in mind, that this method is less prone to error propagation and allows for flexibility in choosing grain shapes (including for instance non-equiaxial grains), it seems to be better suited for the needs of 3-D grain size description in ultrafine-grained materials.