Thermal imaging and spectral data are potential tools that can be used to determine plant water stress. The objectives of this study were to combine thermal and spectral indices to determine water stress, and to estimate yield and investigated traits via these indices for green bean (Phaseolus vulgaris L. cv. Gina) under different water stress levels using stepwise multiple linear regression (SMLR) analysis. The performance of various reference surfaces for Crop Water Stress Index (CWSI) computation was also tested. A field experiment was conducted in Canakkale, Turkey in 2012 consisting of four irrigation treatments, full irrigation (unstressed) and three stress levels applying 25%, 50%, and 75% of full irrigation. To conclude, it was found that the green bean plant was very susceptible to water stress. In determining water stress, Water Band Index (WBI), Normalized Difference Vegetation Index (NDVI), Soil-Adjusted Vegetation Index (SAVI) and Simple Ratio (SR) from the spectral indices tested were found to be highly useful in the discrimination of all treatments. CWSI based on the artificial reference (CWSIa) from thermal indices performed well in discrimination of water stress. Statistical differences were found between wet and dry reference surfaces. Considering both accuracy and ease of measurement in determining references, the use of thermal indices based on artificial references and empirical methods was found to be more useful. The highest R-2 value between yield, investigated traits and indices was between chlorophyll readings (ChIR) and indices with 0.82 according to SMLR. As a result, it can be said that the combined use of thermal and spectral indices can be used as a potential indicator of water stress in green bean.