Abstract. Inspired by a previous study showing a striking geographical mortality clustering, not attributable to chance, in the south-western region of Spain in 1987-1995, the authors have conducted an ecological study of time trends in all-cause mortality risk between 1987-1995 and 1996-2004 in 2,218 small areas of Spain. To identify high-risk areas, age-adjusted relative risks for each area, stratified by sex and time period, were computed using a Bayesian approach. Areas of high-risk in both periods, or in the second period only, were identified. Annual excess mortality and percentage of people living in these high-risk areas, again stratified by sex and time period, were estimated. The cluster of high mortality risk areas identified in the southwest of Spain during 1987Spain during -1995Spain during persisted in the period 1996Spain during -2004 with an increase in the number of highrisk areas and in annual excess of mortality. These increases showed a gender difference, being more pronounced in women.