In the reliability analysis involving fatigue life, statistical information of the geometric parameters and the S-N parameters of the material are necessary to calculate failure probability. Recently, Dimension Reduction Method combined with Kriging approximation (named as K-DR) was developed by the authors, which is an efficient means to construct probability distribution for a response function due to the random input parameters based on the concept of additive decomposition. If all the probability distributions of input parameters are well established in the fatigue life analysis, the K-DR method can be ordinarily employed to obtain PDF of fatigue life. The probability distributions of S-N parameters, however, are not always available due to the limited experimental data. In this case, a family of curves representing confidence bound which is called P-S-N curve are more useful in the design practice. Then the random S-N parameters are turned into a number of deterministic values with different confidence level, at which the PDF’s can be obtained respectively by repeated implementation of K-DR method. By exploiting the concept of additiveness, however, the repetition of K-DR can be avoided, i.e., once an ordinary problem under the random S-N parameters is solved using the K-DR, comparable accuracy can be achieved without implementing new K-DR for the additional problems at each of the deterministic S-N parameters. The proposed method is demonstrated for the fatigue design problem of knuckle. The resulting information is of great practical value and will be very helpful for the design engineers in making decision in the fatigue design process.