A large borehole cavity designed to investigate agricultural chemical transport through the vadose zone was installed at the Agronomy-Agricultural Engineering Research Center near Ames, Iowa. The system consisted of a culvert installed vertically in the soil profile with stainless steel suction lysimeters installed radially from the cavity wall. Suction lysimeters were installed at five depths in the soil profile with five replications at each depth to monitor changes in water quality in the vadose zone and below the water table. Electronic tensiometers and a data logger system were installed to continuously monitor matric potential and used to automatically collect water samples during a leaching event;A rainfall simulation experiment was conducted to test the system and to determine tracer movement in the soil profile. Chloride was dissolved in the sprinkler water to illustrate surface applied water infiltrating through the profile while bromide was applied to the soil surface to simulate soil nitrate movement. Metolachlor and metribuzin were applied three weeks prior to conducting the experiment. Chloride and bromide breakthrough curves showed that these conservative tracers were first detected at the 300 cm depth within 3 h of initiating the 1.5 cm h-1 simulation. Total rainfall application depths ranged from 10.2 to 13.9 cm around the perimeter of the borehole cavity;Mass balances were estimated based on the observed concentrations and showed that from 10 to 30% of the applied chloride tracer was leached below the 180 cm depth and from 4 to 17% of the applied bromide tracer was leached below the 180 cm depth. Applied at 175 kg ha-1, bromide concentrations below the water table increased from zero to approximately 10 mg L-1 which is the maximum contaminant level (MCL) set by the EPA for nitrate-nitrogen. Herbicide concentrations were mostly below detectable levels during the course of the experiment;A two-dimensional convection-dispersion based computer model was used to predict solute transport during the experiment. The model accurately predicted water table response but underestimated solute transport. The observed data showed that preferential flow contributed to solute transport and convection-dispersion based models may not adequately predict solute transport in glacial till soils.