The effect of tillage practices on leaching and persistence of atrazine and metolachlor was evaluated in a field study in the Coastal Plain region of Virginia. Field data were also used to validate pesticide transport models, GLEAMS and PRZM. The study was conducted on two 18x27 m plots located in a field that was in the second year of a two-year no-till wheat-beaDs-com rotation. One plot was conventionally tilled using a moldboard plow and a disk harrow before planting of com and application of chemicals. Soil samples were collected on six sampling dates during the crop growing season at 20 randomly selected locations in each plot with the 0-150 cm sampling depth divided into eight increments. Bromide concentrations were analyzed to provide an estimate of solute movement. High rainfall following chemical application led to rapid leaching of bromide, with the chemical moving faster in the no-till profile. Pesticide concentrations also showed a greater potential for leaching in the no-till plot in the early stages of the study. Chemical concentrations were higher in the no-till profile initially, and were higher in the tilled profile toward the end of the season. Atrazine dissipation was higher in the no-till plot, but there was no marked difference in metolachlor dissipation between the two tillage treatments. Over 35% of atrazine mass remained in the soil profile in both plots at the end of the crop growing season. Pesticide concentrations were found to vary largely over the two plots. The field data were used to evaluate the ability of the pesticide transport models, GLEAMS and PRZM, to represent chemical concentration distribution, depth of solute center of mass, and pesticide mass in the no-till and the conventionally-tilled root zone. The models were evaluated in three sequential steps. The fast simulation was completely uncalibrated, using best available estimates for the input parameters. For the second simulation hydrology parameters were calibrated to minimize errors in the hydrology component so as to better evaluate the prediction of pesticide behavior in soil. The third stage of the evaluation used pesticide dissipation half-life calculated from the field data. Model performance was evaluated using both objective and subjective criteria. GLEAMS and PRZM predicted pesticide concentration in soils reasonably well when run without any calibration. Bromide concentrations were predicted closer to the observed values than pesticides. Overall predictions by both models were better in the conventional tillage plot than in the no-till plot. The comparative effect of tillage on observed chemical concentrations was represented better by GLEAMS than by PRZM. The models under-predicted leaching of pesticides in the early sampling dates. Predicted pesticide mass in the root zone were reasonably close to the field measured values. Calibration of the hydrology component of the models did not improve the prediction of pesticide behavior in soils. The use of field pesticide half-life resulted in better prediction of pesticide persistence but did not improve the overall prediction of pesticide behavior in the two plots. The study identifies selection of input parameters and correct interpretation of results as important factors in the effective use of GLEAMS and PRZM as management tools. / Master of Science