The objective of this work was to evaluate, through a predictive model, which factors influence soybean yield, using agronomic aspects and vegetative indices, in addition to identifying the best soybean cultivar for the northwest of Rio Grande do Sul. The experimental design used was strips with randomized blocks, consisting of 10 cultivars and five blocks. Analyzing the quantitative characters, the positive contributions to yield came from the grain weight of vegetables with two grains, and the grain weight of the plant, having a strong influence on the average grain yield. The vegetable grain weight with three grains contributed negatively to the yield, as the average grain weight was lower than expected, lowering the average grain yield. For the vegetation indices that contributed positively, the GRAY, IGB and RGRI index stand out, while the BGI, GLI2, GRAY2, IGR, IRB, NRBDI and NG indices had negative contributions to the average grain yield. The NEO 581 E cultivar showed better yield performance, reaching 5780 kg ha⁻1, followed by the SOYTECH 541 I2X cultivar, which reached a yield of 5356 kg ha⁻1. The predictive model identified the main variables that influenced final yield, with Cercospora sojina and Corynespora casiicola, grain weight in three-grain legumes, plant grain weight, GRAY index, IGB, NGBDI and RGRI, the variables that contributed positively to grain yield.