This paper analyzes the performance of a fiber Bragg grating (FBG) filter-based strain and/or temperature sensing system based on a modified Gaussian function (MGF) approximation method. Instead of using a conventional Gaussian function, we propose the MGF, which can capture the characteristics of the sidelobes of the reflected spectrum, to model the FBG sensor and filter. We experimentally demonstrate that, by considering the contributions of the sidelobes with the MGF approximation method, behaviors of the FBG filter-based FBG displacement and/or temperature sensing system can be predicted more accurately. The predicted behaviors include the saturation, the sensitivity, the sensing range, and the optimal initial Bragg wavelengths of the FBG sensing system.