Due to the complexity of sedimentary process of the strata, the distribution of porosity in clastic rock is not homogeneous but random. From the point of microscopic structure, pores in reservoir rock are fractal, and the fractal dimension can describe quantitatively the complexity and the heterogeneity of pore distributions. Geographic Information System (GIS) is a kind of technological system collecting, storing, managing, computing, analyzing, displaying and describing the geospatial information supported by computer software and hardware. In the last 15-20 years, GIS have been increasingly used to address a wide variety of geoscience problems. This paper first calculated the pore fractal dimension in three casting thin sections of clastic rock which came from cores of FengTan1 Well and FengCan3 Well in Dangfeng-Xinghua, JiangSu Province using box-counting method, then predicted the porosity according to the porosity fractal model of rocks in GIS environment. In the last we compared the fractal porosity prediction results with mercury experimental data, and we saw the differences were not significant, so the method in this paper to predict porosity of clastic rock is easy to implement and has high precision.