The interpretation of macroseismic questionnaire data is performed by means of a statistical approach: multivariate (correspondence analysis) and trend analysis. The method allows us to objectively estimate the intensity, by evaluating the degree of overlap of different macroseismic effects and by assessing the uncertainty associated with the questionnaires. Estimation of the regional component of the intensity field, subsequently allows us to eliminate, by filtering, local variations in the observations. A criteria for selecting appropriate filter parameters is given. We then apply the method to the questionnaires relative to an Italian earthquake.