Soil salinization is a major factor limiting crop productivity in irrigated agricultural systems under arid climatic conditions. This study aims to develop an early prediction method for maize yield grown on saline soils in the Shaulder irrigated area located in the middle reaches of the Syr Darya River using Sentinel-2 satellite imagery. Field investigations were conducted during three phenological stages of maize development (spring, summer, and autumn) during the 2021 growing season. Soil salinity (electrical conductivity) was measured at 245 observation points in the 0–20, 20–50, and 50–100 cm soil layers, while maize biomass and yield were determined through field-based measurements. Linear regression analysis was used to evaluate relationships between spectral indices and field observations. Among the tested indices, NDVI, GNDVI, OSAVI, and SI7 showed the strongest relationships with soil salinity (R² = 0.76–0.77), while NDVI demonstrated a strong correlation with maize biomass (R² = 0.71–0.74). Time-series analysis of NDVI indicated that soil salinity delays maize development and shifts the timing of peak greenness. On non-saline soils, maximum NDVI occurred on 28 July, whereas on moderately and highly saline soils peak values were observed in early September. Based on peak NDVI values and field-measured yield data, a yield prediction algorithm was developed that enables maize yield forecasting approximately 2–2.5 months before harvest. Estimated yields exceeded 14.2 t/ha on non-saline soils, ranged from 6.8–14.2 t/ha on slightly saline soils, 3.2–6.8 t/ha on moderately saline soils, and 0–3.1 t/ha on highly saline soils. The proposed approach can support improved crop management and irrigation planning in saline agricultural landscapes.