This paper proposes an integrated analysis method for linerless composite vessels considering matrix-crack-induced stiffness degradation and permeation. Stiffness degradation models are obtained using machine learning regression, the data sets of which consist of finite element analysis (FEA) results and virtual samples. Then, the effective leakage conductance is obtained by using the linear elastic fracture mechanics and the micro-cracking leakage mechanism. The proposed integrated method is applied to analyze vessel structures designed by the grid theory, which has variable ply thicknesses and layup angles. Simulation results show that whether the structure at cryogenic temperature cracks and leaks more easily depends on its dimensions and stacking sequences.