This paper presents a control algorithm for blending systems. Such systems are used in refining to produce mixtures having specified properties from several components. The underlying control problem is multi-variable, with constraints on the inputs and outputs, and involves large uncertainties. To address this complexity, a constrained optimization problem is formulated, while the uncertainties are treated in closed-loop by an estimator of the components properties. Besides a theoretical study of the main algorithm and a study of convergence, the paper presents numerous technical details that are needed to solve blending problems as they appear in refining operations. Among them are a general lack of measurements, variable delays, measurement synchronization, infeasibilities management, hydraulic constraints, and pre-blends. Industrial case-studies are provided and stress the relevance of the approach. (C) 2010 Elsevier Ltd. All rights reserved.