The capacity of a ground-based canopy sensor to detect stress-related parameters of cotton (Grossypium hirsutum) was investigated in a split-plot field experiment for two consecutive growing seasons in central Greece. Three levels of irrigation (22, 31 and 40 cm water) were the whole-plot factor and three rates of fertilizer (60, 110 or 160 kg N ha−1) were the split-plot factor with three replications. Irrigation level was the major factor that explained variations in leaf isotopic composition (δ15N and δ13C) within growing seasons and cotton yield at harvest. The rate of fertilizer application did not have a significant effect on cotton yield because there was sufficient residual soil N to meet crop needs. Canopy NDVI was highly correlated to yield when cotton response to differential irrigation was detected. The obtained correlations between canopy reflectance and stress-related parameters (leaf N, δ15N and δ13C) and the stability of the relationship between NDVI and yield over two consecutive seasons indicated that ground-based remote sensing can be used to assess the level of water stress that occurred during the growing season. The application of this technology for in-field monitoring of water stress may prove valuable in semiarid regions where water is often the most limiting factor in crop production.