MEDEA3 was a coordinated project whose main objective consisted in the study of the geographical variability of mortality in (26) large Spanish cities. The units of analysis in MEDEA3 were the census tracts, which is the reason behind the small areas estimation problem. Despite its markedly applied nature, and oriented towards Public Health, the statistical methodology used in MEDEA3 is quite sophisticated. In fact, novel methodological developments were proposed within the project. In this talk we will present the multivariate spatial analysis methodology used in the project, which shares information not only between neighboring regions but also between the different causes of death analyzed. In addition, we will also review the methodological advances, in terms of disease mapping, developed and incorporated into the analyzes of the project itself. Finally, the talk will conclude with a brief summary of the main results of the project, as an example of conclusions that can be obtained in this type of study.