Accurate estimation of wheat biomass plays a crucial role in farm monitoring and management, crop yield prediction, and water use efficiency. Light Use Efficiency (LUE)-based models are widely recognized as an effective approach for biomass estimation due to their physiological basis and their ability to integrate remote sensing and meteorological data. The objective of this study was to evaluate and compare the performance of four vegetation indices—NDVI, NDRI, SAVI, and MSAVI—within the LUE model for estimating wheat biomass in West Azerbaijan Province, Iran. Sentinel-2 satellite imagery, ERA5-Land climate data, and field-measured biomass observations were used in the analysis. Field measurements of biomass were conducted in 30 farms during the 1401 growing season. Model performance was assessed using several statistical evaluation metrics. The results showed that the NDVI-based model achieved the highest accuracy and stability in both the training and validation stages among the evaluated vegetation indices, exhibiting the lowest RMSE and nRMSE values and the highest correlation coefficient and Nash–Sutcliffe efficiency. Furthermore, the Wilcoxon signed-rank test indicated that the performance differences between NDVI and both SAVI and MSAVI were statistically significant, whereas the difference between NDVI and NDRI was not statistically significant. These findings suggest that, under the climatic conditions of the study area, the use of more complex vegetation indices does not necessarily result in a significant improvement in the accuracy of the LUE model.