The invention discloses an early-warning method for judging abnormal furnace conditions of a blast furnace according to static pressure differences. According to changes in the static pressure differences of the blast furnace at different heights in various directions and the overall pressure difference of the blast furnace, airflow distribution and stock column motion in the blast furnace are recognized in time by a data mining technology so as to deduce the time and extents of the changes influencing the furnace conditions, and then the furnace conditions can be timely recovered to a normal state by adopting effective measures so as to guarantee the stability and smooth running of the blast furnace. Through adoption of the early-warning method, early warnings of burden hanging and channeling are given according to the changes in the static pressure differences of the blast furnace, the positions of the abnormal furnace conditions are analyzed, and pertinent measures are adopted for treatment in advance, so that an upcoming abnormal furnace condition can be eliminated or the accident influence can be reduced to the smallest extent.