Abstract: The mires in northern Great Khing'an mountains are particular with high latitudes and high altitudes.A Logistic regression model is developed to analyze the relationship between mire distributions and 18 eco-factors (including 11 topographic and 7 climate factors).The future potential distributions of mires are predicted under CGCM3,and the landscape changes are calculated by Fragstats software.The Logistic model has high prediction precision (ROC=0.96).The prediction results show that:by 2100,under the SRES B1,the potential mire area will decrease about 54.16%,and mires will disappear in the south of the flat hilly regions and intermontane plains; under the SRES A1B,the potential mire area will decrease about 59.62%,and mires in south forest regions will disappear; under the SRES A2,the potential mire area will decrease about 73.51%,and mires will be found only in the north area with higher elevations.In addition,the mean patch areas,landscape shape indexes and aggregation indexes will reduce,which means that the mire distribution will shrink from edges to centers and tend to be fragmented with simple patch shapes.