Different modes of preparation of a by-product of a magnesite calcination process for magnesium oxide production (BMP) were compared for phosphorus removal to struvite at a molar Mg:P ratio of 1.6 mol Mg/mol P from liquor of a full-scale sewage treatment plant. Five experimental runs were performed-Run 1: raw BMP; Run 2: screened to particle size < 0.04 mm; Run 3: milled to particle size < 0.04 mm; Run 4: suspension at a concentration of 3 g of BMPA of tap water; Run 5: supernatant after settling for 4 h after 28 h aeration. Temperature and pH were determined in liquid samples and analyses of phosphorus (P) and magnesium (Mg) were also performed in liquid and solid samples. pH increased with the reaction time (t) in Runs 1-4. However, in Run 5 there was an initial increase from 8.28 to 8.41 only to decrease to 8.34 at the end of the experiment. Phosphorus concentration decreased with t, but also with the mode of BMP preparation where the concentration of phosphorus in relation to the reaction time was lower in Runs 2 and 3 than in the other runs, the minimum being in Run 5. It was found that the smaller the particle size, the lower the concentration remained in solution. This fact was confirmed by the values obtained in the equilibrium concentrations which were: 14.7, 9.1, 13.21, 17.4 and 7.3 mg/l for Runs 1-5, respectively. The amount of struvite obtained was improved by the increase of t, the maximum being in Run 5 at 0.5 h (450 mgA). In Runs 2 and 3, values of struvite production of 400 and 350 mgA were achieved at t higher than 2.5 h. It was found that a first-order kinetic model fitted in well with the experimental data obtained. The values of the reaction constants were found to be: 1.2, 1.5, 1.4, 1.3 and 2.9 h(-1) for Runs 1-5, respectively. The proposed model accurately predicted the performance of the process, showing deviations lower than 5% between the experimental and theoretical data of struvite production. (c) 2007 Elsevier B.V. All rights reserved.