Botanical composition of grass‐legume mixtures greatly influences the productivity and quality of the sward and is therefore an important variable in many agronomic studies. Four indirect methods of estimating the botanical composition of mixed swards of alfalfa (Medicago sativa L.) and smooth bromegrass (Bromus inermis Leyss.) were evaluated to determine their relative efficacy. The methods were the constituent differential method using either neutral detergent fiber (NDF) concentration or crude protein (CP) concentration as variables, a modified constituent differential method where NDF and CP concentrations were used simultaneously as variables, and near infrared reflectance spectroscopy (NIRS). Mean deviations of predicted alfalfa percentages from known values were lowest for NIRS at all stages of maturity and averaged 1.5 percentage units. Of the constituent differential procedures, NDF was the most reliable variable for predicting alfalfa percentage over all maturities with deviations averaging 2.9 percentage units. Based upon the results of this study, NIRS would be the preferred method for estimating botanical composition of grass‐legume mixtures; however, in cases where NIRS is unavailable or inappropriate, the constituent differential method using NDF as a single variable would be an acceptable alternative.