Shrinkage porosity exists more or less in heavy castings, and it plays an important role in the fatigue behavior of cast materials. In this study, fatigue tests were carried out on the QT400-18 cast iron specimens containing random degrees of shrinkage porosity defect. Experimental results showed that the order of magnitude of life scattered from 103 to 106 cycles when the shrinkage percentage ranged from 0.67% to 5.91%. SEM analyses were carried out on the shrinkage porosity region. The inter-granular discontinuous, micro cracks and inclusions interfered with the fatigue sliding or hindering process. The slip in shrinkage porosity region was not as orderly as the ordinary continuous medium. The shrinkage porosity area on fracture surface(SPAFS) and alternating stress intensity factor(ASIF) were applied to evaluate the tendency of residual life distribution; their relationship was fitted by negative exponent functions. Based on the intermediate variable of ASIF, a fatigue life prediction model of nodular cast iron containing shrinkage porosity defects was established. The modeling prediction was in agreement with the experimental results. Shrinkage porosity exists more or less in heavy castings, and it plays an important role in the fatigue behavior of cast materials. In this study, fatigue tests were carried out on the QT400-18 cast iron specimens containing random degrees of shrinkage porosity defect. Results showed that the order of magnitude of life scattered from 103 to 106 cycles when the shrinkage percentage ranged from 0.67% to 5.91%. SEM analyzes were carried out on the shrinkage porosity region. The inter-granular discontinuous, micro cracks and inclusions interfered with the slipage porosity area on fracture surface (SPAFS) and alternating stress intensity factor (ASIF) were applied to evaluate the tendency of residual life distribution; their relationship was fitted by negative exponent functions. Based on the intermediate variable of ASIF, a fatigue life predicti on model of nodular cast iron containing shrinkage porosity defects was established. The modeling prediction was in agreement with the experimental results.