A great deal of interest has been directed toward the development and application of crop simulation models and accompanying artificial intelligence programs. GOSSYM/COMAX is a crop management system developed by the USDA-ARS Crop Simulation Laboratory for cotton (Gossypium spp.) which utilizes GOSSYM, a crop simulation model, and COMAX, the artificial intelligence component which exercises GOSSYM and provides management recommnedations in an effort to optimize nitrogen (N), water, and crop termination management decisions. GOSSYM simulates numerous aspects of N transformation pathways in the soil, including leaching and uptake. For GOSSYM/COMAX to operate effectively it is imperitive that GOSSYM is capable of simulating the critical and dynamic aspects of the soil-plant system accurately. To date, very little information is available concerning field testing and validation of GOSSYM, particularly with respect to the fate and distribution of soil NO3--N. If N is to be managed for optimum efficiencies from an agronomic standpoint, as well as within sound environmental management constraints, N fertilizer efficiencies must be optimized. Management of N is particularly critical in an irrigated cotton production system, and GOSSYM has been offered as an effective method of managing fertilizer N inputs. The objective of this study was to compare measured soil NO3--N concentrations and cotton response within an irrigated cotton production system to those simulated by GOSSYM. Soil and plant data was collected from a field of Upland cotton (G. hirsutum L., var. DPL 90) on a Mohall sandy loam soil (fine-loamy mixed, hyperthermic Typic Haplargid) near Casa Grande, AZ. All necessary soil, crop, weather, and management input information required to conduct valid GOSSYM simulations were provided. Details associated with the field methodology employed is described in a companion paper (Silvertooth et al., 1992. SSSAJ 56:548-555). Soils were sampled from a mainplot and subplot configuration within a uniformly managed field in a manner which allowed for a description of soil NO3--N distibution patterns both spatially and temporally. Soils were sampled at 30 cm depth increments to a total depth of 180 cm at five seperate dates over the course of the season. Soil sample data revealed a high degree of spatial and temporal variability. GOSSYM's simulations of soil NO3--N concentrations always fell within the standard deviations of the mean for the measured values, but trends were such that GOSSYM tended to overestimate NO3--N leaching through the profile, indicating a high degree of sensitivity to irrigation water inputs. A comparison of simple soil-crop N budgets for field and simulated cases were very close. However, GOSSYM overestimated yields by 30%, where measured yield was 1,680 lb. lint acre-1. These results indicate the need for a further refinement of the mechanisms within the model which govern fertilizer N transformations, leaching, and crop uptake and response, which will be particularly important under irrigated conditions.