It is much more significant for assessment of probability distribution of large samples of geotechnical parameters to study the stability and reliability of geotechnical engineering in some important geotechnical engineerings. Thus normal information spread estimation method (NISEM) is proposed to generate probability density functions of geotechnical parameters with large samples based on the optimal bandwidth which is received by the minimum mean integrated square error. The method proposed in this paper based on the information spread principle

instead of assuming the fitting test of classical probability distribution curves

makes full use of the data sample provided from the view of the test sample and the information theory. It presents the mathematical and physical performances more fully and tight. The influence of bandwidth on the results of information diffusion estimation has been studied by the inference of probability density function of compression index (Cc). The results show the reasonability of NISEM

which is used to infer probability density function of geotechnical parameters with large samples. At last

probability density functions of rock and soil shear strength parameters have been inferred by using NISEM. By K-S test

the correctness and practicability of NISEM are verified.