The recent fire history of northern Quebec biomes (54 000 km2), including the northern Boreal Forest, the southern and northern Forest—Tundra, and the Shrub Tundra, was documented by examining size and dates of 20th century wildfires using tree ring techniques. Results showed that pronounced south—north differences in fire properties existed, corresponding to climate and vegetation gradients. Fire frequency per biome decreased south—north from closed forest (0.7 fire/yr) to Shrub Tundra (0.4 fire/yr). Average fire size decreased south—north by 100—fold from °8000 ha in the northern Boreal Forest to 80 ha in the Shrub Tundra, while modal fire size was <50 ha in each of the four biomes. Most fires (>80%) of the northern Forest—Tundra and the Shrub Tundra were <100 ha, and fires >100 000 ha occurred only in the northern Boreal Forest and the southern Forest—Tundra. Less than 35% of all mapped fires in the Boreal Forest were <50 ha, but >30% were >1000 ha. From south to north, the fire—free interval per biome was, respectively, °2.6, 0.6, 0.6, and 2.2 yr, the Boreal Forest data being overestimated. The largest burned areas were recorded in the 1950s throughout the biomes, most likely associated with longlasting drier and warmer conditions. The fire rotation period per biome, based on the percentage of burned areas during the 1920—1984 period (or 1930—1984 in Tundra), increased south—north by 100—fold from 100 yr in the northern Boreal Forest to 9320 yr in the Shrub Tundra. The fire rotation period around the tree line, i.e., 20 km south and north of the present tree line, was estimated to be >7800 yr. Biome boundaries have developed and are maintained in response to fire by the ability of spruce to seed and regenerate. Stability of northernmost conifer sites is maintained by (1) the inability of patchy shrub and conifer cover in the northern Forest—Tundra and Shrub Tundra to carry fire and (2) failure of trees to produce viable seeds in these two biomes. Present data suggest that the area is characterized by a much higher fire frequency than expected from the fire weather index and from calculated frequencies typical of vegetation—type studies. It is concluded that size of the study area is a key element in the determination of regional fire regimes. Finally, the ecological significance of the natural fire rotation and postfire regeneration in northern environments is discussed in a paleoecological perspective.