To study the characteristic of soil-water characteristic curves (SWCC) of several different types of soils, a number of data sets of matric suction-volume water content were studied for UNSODA clay, UNSODA sand, Guangxi expansive soil and Hefei expansive soil. Three types of Fredlund-Xing models and three types of van Genuchten models were used to study the influence of SWCC models, and three types of functions in the scientific programming language MATLAB were applied to analyze the effect of different fitting methods. Through the data fitting and statistic analysis, the probabilistic characteristics of the fitting parameters of SWCCs for the four types of soils were obtained. For analyzing the relative sensitivity of the normalized volume water content to the three fitting parameters of SWCC, a formula which considers the correlation of parameters was proposed. Through the data fitting of many data sets, it is found that the fitting methods influence the value of parameters of SWCC, while the influence is less than that of fitting models from the statistical point of view. The SWCCs of all the models are nearly the same when the suction is low. With the increase of the suction, the SWCCs of FX2 and VG2 are very close, and the SWCCs of FX3 and VG3 are the same. The sensitivity analysis shows that the normalized volume water content is more sensitive to parameters n and m than to parameter a.