Shale oil reservoirs, characterized by the complex occurrence state of fluid and complex mineral compositions, are difficult to accurately determine the fluid component content. Therefore, a method to determine the fluid component content of shale oil combining with continuous wavelet transform (CWT) and Gaussian distribution function is proposed in this paper. Based on the focused ion beam scanning electron microscopy experiments, a digital shale core containing inorganic pore water, organic pore oil, matrix, pyrite, and organic matter is obtained. On this basis, the random walk method is used to simulate the nuclear magnetic resonance response of the pore fluids at different echo spacing. The echo data is inversed for the T2 distributions of inorganic pore water, organic pore oil, and organic matter as well as the T2 distribution of all fluid components. For the T2 distribution of all fluid components, the CWT is carried out, and a two-dimensional wavelet coefficient matrix can be obtained. By identifying spectral peaks, the positions and shape parameters of spectral peaks at different scales in the wavelet space are determined. The T2 distribution for all fluid components is decomposed by the constructed Gaussian distribution functions of different scales. The optimal decomposed result is used to calculate the fluid content of shale oil. The contents of fluid components calculated by proposed method with the numerical simulation based on random walk method at different echo spacing are compared. The result indicates that the absolute error between them is relatively low at different echo spacing. The effectiveness of the proposed method is verified. This method can provide technical support for the exploration and development of shale oil.