In a part of the municipal territory of Reggio Emilia, northern Italy, selenium in drinking water decreased from 7 micrograms/L to less than 1 micrograms/L. In a cohort of 4419 individuals, previously exposed for at least 5 yr to the drinking water with higher selenium content, the 7-yr temporal distribution of deaths for coronary disease and for stroke was analyzed to examine a possible relationship with changes in drinking water selenium. From January 1986 until August 1988, when tap water selenium was 7 micrograms/L, deaths for coronary disease were one in males and two in females. After the decrease in drinking water selenium, 21 and 10 coronary deaths were observed, respectively, in males and in females from September 1988 to December 1992. No significant difference in the temporal distribution of stroke deaths was observed both in males and in females. Even if an effect of chance and aging in the temporal distribution of coronary deaths may not be excluded, findings of the study seem to be consistent with the hypothesis of a beneficial effect of selenium on coronary disease mortality.