Understanding plant interactions is greatly limited by our ability to identify and quantify roots belonging to different species. We proposed and compared two methods for estimating the root biomass proportion of each species in artificial mixtures: near-infrared reflectance spectroscopy (NIRS) and plant wax markers. Two sets of artificial root mixtures composed of two or three herbaceous species were prepared. The proportion of root material of each species in mixtures was estimated from NIRS spectral data (i) and the concentration patterns of n-alkanes (ii), n-alcohols 10 (iii), and n-alkanes + n-alcohols combined (iv). For each data set, calibration equations were developed using multivariate statistical models. Botanical composition of root mixtures was predicted well for all the species 13 considered. The accuracy varied slightly among methods: alkanes < alcohols <= alkanes+alcohols < NIRS. Correlation coefficients between predicted and actual root proportions ranged from 0.89 to 0.99 for alkanes + alcohols predictions and from 0.97 to 0.99 for NIRS predictions. These two methods provided promising potential for understanding allocation patterns and competitive interactions.