We used the statistics of extreme values to characterize dry-spell extremes, or runs of consecutive days with little or no rain, and forest fire extremes in various geographical regions in the province of Ontario. Forest fire report records for the 1976-1999 period were used to investigate regional differences in extreme fire event distributions for 10 areas characterized by their unique ecological classification and level of forest fire protection. Precipitation data were used to compare extreme dry-spell event distributions over the 1963-1998 period for 14 weather stations located across the province. Regional comparisons indicated significant differences in the distributions of both extreme fire events and extreme dry-spell events. Non-linear regression was used for maximum likelihood estimation of distribution parameters to calculate return times for extreme fire and dry-spell events. Spearman rank correlations indicated a significant positive correlation between the return times of 100-, 1000-, and 10 000-ha fire events and the return times of 40-day dry-spell events across the province. Comparisons of estimated regression parameters also indicated that the statistical properties of extreme fire events and extreme dry-spell events varied in a similar manner across the province. Results indicated that variation in the magnitude of extreme fire events is related to ecological classification, patterns of extreme dry-spell events, and levels of fire protection.