The occurrence of agglomeration in fluidized-bed combustion and gasification of biomass and waste can have a very negative impact on the efficiency of reactor operation. The formation of agglomerates seriously affects the otherwise uniform heat distribution in the bed and may finally lead to defluidization and unscheduled shutdown of the plant. A recently developed monitoring method is able to detect the onset of such undesired behavior, making it possible to operate more efficiently. This Early Agglomeration Recognition System (EARS) uses a characteristic process variable (e.g., pressure) measured at high frequency. The sensitivity and selectivity of EARS is illustrated with experimental results and the first industrial test results in an 80-MWth wood-fired fluidized-bed combustion plant is discussed. (c) 2006 Elsevier Ltd. All rights reserved.