Although dicamba-resistant crops can provide an effective weed management option, risk of dicamba off-site movement to sensitive crops is a concern. Previous research with indeterminate soybean identified 14 injury criteria associated with dicamba applied at V3/V4 or R1/R2 at 0.6 to 280 g ae ha−1. Injury criteria rated on a 0 to 5 scale (none to severe), along with percent visible injury and plant height reduction, and canopy height collected 7 and 15 d after treatment (DAT) were analyzed using multiple regression with a forward-selection procedure to develop yield prediction models. Variables included in the 15 DAT models (in order of selection) for V3/V4 were lower stem base lesions/cracking, plant height reduction, terminal leaf epinasty, leaf petiole droop, leaf petiole base swelling, and stem epinasty, whereas for R1/R2 variables were lower stem base lesions/cracking, terminal leaf chlorosis, leaf petiole base swelling, stem epinasty, terminal leaf necrosis, and terminal leaf cupping. To validate the models, experiments including the same dicamba rates and application timings used in previous research were conducted at two locations. For the variables specific to each model, data collected for the dicamba rates were used to predict yield. For the V3/V4 15 DAT model, predicted yield reduction (compared with the nontreated control for dicamba at 0.6 to 4.4 g ha−1) underestimated or overestimated observed yield reduction by an average of 1 and 3 percentage points. For 8.8 g ha−1, predicted yield reduction overestimated observed yield reduction by 8 points and for 17.5 g ha−1 by 20 points. For the R1/R2 15 DAT model, predicted yield reduction for 0.6 to 4.4 g ha−1 overestimated observed yield reduction by an average of 3 to 5 percentage points. For dicamba at 8.8 g ha−1, predicted yield reduction underestimated observed yield reduction by 8 points and for 17.5 g ha−1 overestimated by 6 points.