From the data of two climate scenarios 20C3M of 1951-2000 and SRESA1B of 2001-2100 and the data of the 1981-1990 surveys of distributions of red-crowned crane breeding habitats in Northeast China, 5 climatic factors that may affect stay of red-crowned crane communities in their breeding habitats were extracted, using the theory and methods of the normalized-distance-based hierarchical clustering and information fusion technology, for establishment of a mathematical model for stochastic analysis of sojourn time of red-crowned cranes in their breeding habitats and at the same time, algorithm for the model was studied. With the aid of ArcGIS 9.3 and Matlab, programs were designed and run and in the end, a timetable was figured out predicting the most suitable, the second most suitable and suitable sojourn time periods for red-crowned crane communities in Northeast China during the years from 2041 to 2050 and from 2091 to 2100. During the period of 1981-1990, crane communities stayed in the habitats from April to September, while the table listed April, May, August, September and October as suitable sojourn time for all the three grades of suitable sojourn time periods in the years of 2041-2050 and only April, May, September and October in the years of 2091-2100. The comparison shows that the breeding habitats of red-crowned cranes in Northeast China in 1981-1990 will no longer be suitable for red-crowned cranes as breeding habitats in 2041-2050 and 2091-2100, due to climate change. Furthermore, the 3 major temperature-related factors affecting stay of red-crowned crane communities in their breeding habitats are monthly mean of daily maximum temperature, monthly mean temperature and monthly mean of daily minimum temperature, which follow a declining order of monthly mean of daily maximum temperature > monthly mean temperature > monthly mean of daily minimum temperature in extent of their impacts.