Significant climatic changes over Northern Eurasia during the 20th century have been reflected in numerous variables of economic, social, and ecological interest, including the natural frequency of forest fires. For the former USSR, we are now using the Global Daily Climatology Network and a new Global Synoptic Data Network archive, GSDN, created jointly by U.S. National Climatic Data Center and Russian Research Institute for Hydrometeorological Information. Data from these archives (approximately 1500 of them having sufficiently long meteorological time series suitable for participation in our analyses) are employed to estimate systematic changes in indices used in the United States and Russia to assess potential forest fire danger. We use four indices: (1) Keetch-Byram Drought Index, (KBDI; this index was developed and widely used in the United States); (2) Nesterov, (3) Modified Nesterov, and (4) Zhdanko Indices (these indices were developed and widely used in Russia). Analyses show that after calibration, time series of the days with increased potential forest fire danger constructed using each of these three indices (a) are well correlated and (b) deliver similar conclusions about systematic changes in the weather conditions conducive to forest fires. Specifically, over the Eastern half of Northern Eurasia (Siberia and the Russian Far East) statistically significant increases in indices that characterize the weather conditions conducive to forest fires were found. These areas coincide with the areas of most significant warming during the past several decades south of the Arctic Circle. West of the Ural Mountains, the same indices show a steady decrease in the frequency of "dry weather summer days" during the past 60 yr. This study is corroborated with available statistics of forest fires and with observed changes in drought statistics in agricultural regions of Northern Eurasia. (C) 2006 Elsevier B.V. All rights reserved.