The adopted conventional granulation and proposed pre-wetting granulation for six sinter mixtures were reinvestigated and interpreted from the perspective of particle size analysis. To research the experimental data from the two processes, two mass fractal models were adopted and further discussed in the characterization of particle size distribution for these sinter mixtures. It was found that: (1) Compared to conventional granulation, the pre-wetting granulation can significantly change the particle size distributions of the sinter mixtures before and after granulation. (2) A fractal characteristic for the particle sizes of the sinter mixtures was observed in the two granulation processes. The particle size distributions of the mixtures can be simulated by the proposed fractal models. (3) Using such fractal model as a bridge, the granule sizes can be predicted by moisture content and fractal parameters of the researched sinter mixture. However, due to the preferential growth of pre-wetted particles, the predictive models on particle sizes built in conventional granulation did not work well in pre-wetting granulation. (4) A prediction model on green bed permeability was developed with fractal parameters of the granules. With these findings, this work can be meaningful for ore-blending optimization and granulation enhancement.