There are two types of methods for surrogate modeling: (1) Single-step method: By conducting a single design of experiment (DOE) and sampling, the required samples are obtained and used to construct a surrogate model. The number of training samples needs to be determined in advance in the single-step algorithm. If this number is too large, huge computational resources are required to train the samples. Otherwise, it is impossible to guarantee the accuracy of the surrogate model. There is no general method for solving the problem of how to reasonably determine the number of training samples. (2) Sequential method: Firstly, we obtain a small initial set of samples through DOE and construct a surrogate model. Then, the distribution features of the current sampling set and the surrogate model feature are analyzed. After that, new samples are added according to certain strategies and then the surrogate model is updated. The aforementioned steps of adding sequential samples and updating the model shall be repeated until the termination conditions are met (e.g., reaching the predefined accuracy threshold or the upper limit of computational resources). This method can make a reasonable trade-off between the number of training samples and the accuracy of the surrogate model, and create a surrogate model that meets the accuracy requirements at a small cost with an effective sequential sampling strategy. In this chapter, Sect. 6.1 introduces the sequential surrogate modeling method; Sect. 6.2 describes how to obtain samples by preliminary DOE; Sect. 6.3 briefs the sequential sampling and modeling strategy; Sect. 6.4 gives numerical examples for sequential surrogate modeling of explicit functions and implicit functions respectively.