2631ReseaRch Y ield is influenced directly and indirectly by a number of factors, such as plant morphology and physiology, and especially environmental conditions (Prasad et al., 2007a). A good understanding of the factors responsible for growth and development is required to identify an indirect selection tool that can be used to improve yield (Richards, 1982). Reynolds et al. (2001) emphasized the potential of using different morphophysiological selection criteria to complement empirical selection for yield, which potentially can make the selection process more efficient.Digital image analysis is an important tool in biological research and has been applied to satellite images, aerial photographs, and macroscopic and microscopic images. A relevant application of image analysis that has been used for decades is the area of remote sensing. In agricultural and forestry systems, analyses including species identification and estimation of plant cover area and biomass of the above-ground canopy have been conducted using satellite and/or airborne images (Klassen et al., 2003). Image analysis techniques have found a recent application in estimating the biomass of individual plants in controlled ABSTRACT Improving crop productivity in drought-prone environments is a daunting challenge. Selection of advanced breeding materials for yield is a labor-intensive procedure and sometimes produces misleading results because of the complex genetic behavior of yield. remote sensing techniques can provide an instantaneous, nondestructive, and quantitative assessment of a crop's ability to intercept radiation and photosynthesize. The objective of this study was to examine vegetation indices derived from aerial images as biomass and yield prediction tools for soybean [Glycine max (L.) Merr.] under different levels of water availability. Two commercial soybean cultivars with contrasting maturity were planted on a rooting-depth restriction installation. Multispectral aerial images were acquired at early flowering and during seed filling, and fifteen vegetation indices were calculated and their associations with yield and biomass assessed. The indices estimated using the near infrared (NIr), rED, and GrEEN portions of the spectrum were weak predictors of soybean yield under severe water stress conditions. However, under moderate drought or unstressed conditions, the regressions were able to explain up to 80% of the data on the basis of R 2 values. The nominally best relationships with yield were found for NIr from images taken at seed fill and with biomass for rED bands extracted from images taken at flowering. results suggest that aerial imaging shows potential as a tool for yield and biomass prediction of soybean cultivars.