Agglomeration of ash in fluidized bed combustors may result in defluidization and subsequent downtime of the installation. Previous research has shown that Ca-based additives can prevent agglomeration, but the added amount was determined arbitrarily and testing occurred only on lab scale or pilot scale. This paper presents a statistical approach, based on a newly developed agglomeration index, to calculate the amount of CaO that should be added (in the form of a Ca-based mineral, e.g., CaCO3) to the fluidized bed in order to prevent agglomeration. The agglomeration index is based on an understanding of the reactions occurring in the ash, for instance, the formation of potassium silicates with low melting points, and the formation of calcium phosphates and calcium silicates with high melting points. Full-scale testing of partial replacement of silica sand by calcite (CaCO3) as fresh bed material showed that the increased CaO concentration in the ash, with respect to normal operation, appears to reduce both wall and in-bed agglomeration problems. As a measure for agglomeration risk, differential bed pressure variations were statistically analyzed. In the test periods during which CaCO3 was added, the bed pressure variations were smaller and less frequent, and the severity of agglomeration was thus reduced. The proposed strategy can be applied for fuels that are commonly perceived as difficult or unsuited for fluidized bed combustion, and also for other additives than CaCO3, e.g., Al-based minerals.