In Hong Kong, tremendous amount of rock blasting activity has been carried out in recent years, primarily due to the sharp increase in infrastructure projects. Furthermore, it is expected that more blasting work will be undertaken in foreseeable years/decades owing to the continuing development of the mass transit railway system and the use of rock cavern promoting by the government. Traditionally a simple empirical equation in a logarithm-logarithm space with two fitting parameters is employed by engineers to decide the amount of explosive with a given peak ground vibration limit. At the initial stage of a project when limited blast records are available to obtain the site-specific fitting coefficients, reference is often made to the two fitting coefficients proposed by Li and Ng (1992). In our work, a numerical method is proposed to shed light on the subject from a Bayesian probabilistic perspective. Together with Monte Carlo simulation, the development of site-specific fitting coefficients at an early stage of a blasting project is explored. Data from a site formation project is used as an example to illustrate the performance of the method.