The purpose of this work is to build a prediction model of weekly general mortality in Barcelona, using weekly burials as a quick indirect indicator of mortality, and to illustrate its use in the detection of a suspected epidemic of influenza. Mortality data have been modelled following the methodology proposed by Box and Jenkins, and have been regressed using a simple linear model against burial data, so that for every week the expected number of deaths is computed. The model identified that best fitted mortality data was an ARI-MA (3,0,0). The lineal regression model relating mortality and burial data was deaths = 15.94 + 0.80 burials. The use of this methodology allowed the detection of the increase in deaths which occurred during an influenza epidemic in the beginning of 1987. This method allows the rapid weekly monitoring of mortality in a well defined geographic area, and it is a potentially useful tool to support public health departments in their epidemiologic surveillance responsibilities.