High-pressure grinding rolls have an important role in several iron ore pellet feed preparation circuits. Modeling and simulation are potentially useful tools to improve their performance, but the applicability of mathematical models from the literature in this particular application has not yet been demonstrated, although the model by Torres and Casali has been identified as having good potential. The present work relies on several surveys on four HPGRs with variable diameters (1.4 to 2.25 m) and aspect ratios (0.88 to 1.45) operating in pelletizing plants to test the model originally proposed by Torres and Casali, as well as modifications proposed to it. Inadequacies have been found in the original model, which gave poor predictions of power, throughput and product size distribution along the roll's length. Modification of the power equation, incorporating a machine-dependent correction factor for the nip angle, as well as several modifications to the throughput model, including an expression meant to account for the ejection of material from the edge of the rolls and extrusion, resulted in very good agreement between data from surveys and the model. Finally, incorporation of an empirical correction equation to account for saturation and a more general expression for the roll profile led to good agreement be-tween measured and predicted product size distributions, including their variation along the roll's length. ? 2021 Elsevier B.V. All rights reserved.