A field-based spectrometer was used to characterize the spectral reflectance from the cotton canopy at various growth stages (pinhead square to early bloom). A spectral index model was developed based on extracting reflected light energy data in four wavelength bands ranging from 460 to 770 nm. A good correlation (R²=0.91) existed between soil applied nitrogen and cotton yield. Correlations between petiole nitrate and soil applied nitrogen ranged from 0.85 to 0.90 depending of the stage of plant growth. The average correlations between spectral index and petiole nitrate (R²= 0.74) were very promising as a possibility for remotely sensing nitrogen deficiency in cotton at a growth stage where supplemental nitrogen could be applied site-specifically. The implementation of artificial neural networks also showed promise as a possibility to further increase the measurement accuracy by identifying patterns in the spectral data.