This article presents a mathematical model based on linear programming to support decision-making when selecting the soybean variety to be processed for oil extraction and bran production, since the appropriate choice can maximize industry's profits. Samples of nine soybean varieties produced in the town of Dois Vizinhos, in the southwest region of the state of Paraná, Brazil, were used in this study. Laboratory analyses revealed the chemical composition of the grains, the byproducts, residues and losses. The model made it possible to estimate the profits from the processing of each variety and of the nine varieties' average. Our results show that the chemical composition of the grains directly influences the industry's profits and that the processing of the varieties 'BRS133', 'CD215', 'EMBRAPA48', 'BRS184', 'SPRING8350' and 'M-SOY5826' can maximize the profits, while the processing of the varieties 'CD 205', 'CD 206' and 'BRS 214' can reduce them, compared to the average obtained from the investigated varieties.
linear programming; mathematical model; soybean variety