The objective of this study was to evaluate temporal changes of stroke in an Italian community by comparing the present incidence rates with those reported in the same area for 1989. The two studies were conducted by the same research group and met almost all the criteria proposed for an "ideal" stroke incidence study. The annual incidence rate per 1000 inhabitants increased (p < 0.01) by 29%, from 2.23 (95% CL, 1.96-2.50) in 1989 to 2.89 (95% CL, 2.58-3.20) in 1997. No statistically significant change was found when these rates were adjusted to the 1991 Italian population. The overall incidence rate was 2.40 (95% CL, 2.14-2.66) in 1989 and 2.65 (95% CL, 2.39-2.91) in 1997. The thirty-day case fatality rate declined dramatically (p < 0.001) from 31% (95% CL, 26-36) to 20% (95% CL, 16-24) between 1989 and 1997. Ageing of the population and better identification of cases could explain the high incidence rate, whereas the decrease of fatality rate may be due to a general improvement in acute care and inclusion of milder cases.