Below-ground processes of the Gossym cotton simulation bodel were evaluated by comparing simulation output to soil moisture data collected during the 1991 growing season under three replicated irrigation regimes, on both cropped and bare soil conditions. Soil moisture measurements were made weekly using a neutron moisture meter at either three or five position across the crop row and at six depths per position. Moisture content at the soil surface was obtained gravimetrically using a soil sampler of known volume. Wet weather int he spring and regular rainfall throughout the growing season rendered irrigation unnecessary. Irrigation treatment data were pooled and all data classified as reflecting either cropped or bare soil conditions. These data were compared to the output of GOSSYM simulations of the 1991 growing season for cropped and bare soil conditions. The data indicate that Gossym, in this application overestimated soil moisture content by between 5% and 12% at the various depths and row positions. Higher simulated soil moisture content at the soil surface on the bare plots indicates that GOSSYM may underestimate evaporative soil moisture loss. In the cropped plots higher simulated soil moisture at the soil surface may indicate either an underestimation of the amount of shallow subsurface feeding of the crop roots, or an underestimation of evaporation from the soil surface due to shading by the crop canopy. The model appears sensitive to differences between laboratory determined, and actual, moisture content at field capacity. The simulation also exhibited a deep rooted growth habit that was absent in the field. This indicates that the model is unable to accurately simulate root growth when adequate moisture and nutrients are available immediately below the soil surface.