In this paper the authors provide a probabilistic method for the macro-seismic characterization at the site. The probability function for the epicentres location P(x,y|M), assumed as a Gaussian distribution function, is calculated by the main seismic parameters, namely the magnitude, the intensity and the location of the epicentres, for any recorded series of historical events. The final expression of the probability function is set by the addition of the single contributions of n Gaussian functions and is defined by means of the evaluation of the respective unknown parameters (the average values, the variances, the correlation coefficients and the convex combination coefficients) obtained through a ‘fitting procedure’ minimizing the mean square error of the current moments calculated on the basis of any given set of parameters compared to the average moments estimated on the basis of historical data under suitable constraints. In the final part of the paper, this procedure is developed for the probabilistic treatment of the epicentres location in a seismic area in Southern Italy, where earthquakes occurred in a predefined time range with varying magnitudes are considered as input data. Numerical results obtained by the presented simplified methodology show a good agreement, for the specific study-area, with generical seismic assessment presented on larger scale in literature; this demonstrates that the proposed approach ensures a good approximation of the combined probability curves about the epicentral distribution over the considered territory.