The increased use of legume–grass mixtures in grazing systems has resulted in the need for more accurate forage estimation techniques for these diverse mixtures. In mixed stands, variations in species composition and canopy density make the estimation of herbage mass (HM) challenging. Three nondestructive sampling techniques [a rising plate meter (RPM), a pasture ruler, and digital image analysis] were compared with traditional botanical hand separation to determine their effectiveness in predicting HM in bermudagrass [Cynodon dactylon (L.) Pers.] and alfalfa (Medicago sativa L.) pastures when harvested to a 1- or 4-inch stubble height. All models calibrated to forages harvested to 1 inch failed to meet the formal assumptions of the F-test. The height and RPM were moderately successful at predicting HM when calibrated to samples harvested to 4 inches. However, the models were unable to predict HM above 4,000 lb acre−1 in their current form. Future research in alfalfa–bermudagrass mixtures should focus on isolating how HM changes within the canopy and increase the sample area used to collect the calibration data.