An evaluation method collectively reflecting the influence of material tensile strength and specimen size on the failure probability was introduced to accurately characterize the correlation between microstructure and failure probability of cylinder head material. Firstly, based on the results of microstructural characterization and mechanical property testing of different positions in cylinder head, the grey relational analysis method was employed to reveal the relationship between microstructural characteristics and mechanical properties by using two-dimensional porosity, pore area, pore maximum Feret size, silicon phase area, silicon phase length-width ratio, silicon phase maximum Feret size, silicon phase particle roundness as relevant sequences and using tensile strength, elastic modulus, Vickers hardness as reference sequences. Based on the weakest chain theory, a generalized two-parameter Weibull distribution model with weakened specimen size effect was built to describe the relationship between failure probability and material strength at different locations of cylinder head by considering the degree of defect distribution deviation from the ideal uniform distribution. The results showed that the pore maximum Feret size had a relatively large role in influencing the mechanical properties of cast aluminum cylinder head body material, the bottom plate position was prone to the centralized distribution of local defects, and the probability of failure caused by casting defects was higher.