Fatigue life consists of two parts: crack initiation life and crack propagation life. For weld compo-nents it is inevitable that defects exist. These defects form“notch”in the weld and increase the local tensile stress, which is called “notch effect”. Notch effect makes crack initiation life very short, so fa-tigue life of weld components approximately is equal to crack propagation life. These defects in the weld can be regarded as “initiation crack”. The uncertainty of crack propagation rate mainly comes from weld initial crack size, material properties and applied loading in the pressure hull fatigue life reliability prediction. Welding process, nondestructive inspection error and so on may cause uncertainty of weld initial crack size. Due to the dispersed nature of material mechanical property and uncertainty of the specimen test, material parameters are usually treated as random variables. Hydrostatic pressure is the main loading for submarine pressure hull. Its variation caused by submarinefloating and sub-merging will lead to fatigue damage in the pressure hull weld. But the submerging depth is random every time, consequently the hydrostatic pressure is random, so mathematical statistical method is used here. The submarine submerges to a certain depth,finally itfloats to the surface, which is defined as a loading cycle. In this definition loading amplitude is maximum, which is a partial safety from the viewpoint of fatigue damage. Besides, loading cycle counting can be achieved through the statistics of submarine submergence depth, which is very convenient.