The movement of water and tritium through the unsaturated zone was studied at a low-level radioactive-waste disposal site near Sheffield, Bureau County, Illinois, from 1981 to 1985. Water and tritium movement occurred in an annual, seasonally timed cycle; recharge to the saturated zone generally occurred in the spring and early summer. Mean annual precipitation (1982-85) was 871 millimeters; mean annual recharge to the disposal trenches (July 1982 through June 1984) was estimated to be 107 millimeters. Average annual tritium flux below the study trenches was estimated to be 3.4 millicuries per year. Site geology, climate, and waste-disposal practices influenced the spatial and temporal variability of water and tritium movement. Of the components of the water budget, evapotranspiration contributed most to the temporal variability of water and tritium movement. Disposal trenches are constructed in complexly layered glacial and postglacial deposits that average 17 meters in thickness and overlie a thick sequence of Pennsylvanian shale. The horizontal saturated hydraulic conductivity of the clayey-silt to sand-sized glacial and postglacial deposits ranges from 4.8x10^-1 to 3.4x10^4 millimeters per day. A 120-meter-long horizontal tunnel provided access for hydrologic measurements and collection of sediment and water samples from the unsaturated and saturated geologic deposits below four disposal trenches. Trench-cover and subtrench deposits were monitored with soil-moisture tensiometers, vacuum and gravity lysimeters, piezometers, and a nuclear soil-moisture gage. A cross-sectional, numerical ground-water-flow model was used to simulate water movement in the variably saturated geologic deposits in the tunnel area. Concurrent studies at the site provided water-budget data for estimating recharge to the disposal trenches. Vertical water movement directly above the trenches was impeded by a zone of compaction within the clayey-silt trench covers. Water entered the trenches primarily at the trench edges where the compacted zone was absent and the cover was relatively thin. Collapse holes in the trench covers that resulted from inadequate compaction of wastes within the trenches provided additional preferential pathways for surface-water drainage into the trenches; drainage into one collapse hole during a rainstorm was estimated to be 1,700 liters. Till deposits near trench bases induced lateral water and tritium movement. Limited temporal variation in water movement and small flow gradients (relative to the till deposits) were detected in the unsaturated subtrench sand deposit; maximum gradients during the spring recharge period averaged 1.62 millimeters per millimeter. Time-of-travel of water moving from the trench covers to below the trenches was estimated to be as rapid as 41 days (assuming individual water molecules move this distance in one recharge cycle). Tritium concentrations in water from the unsaturated zone ranged from 200 (background) to 10,000,000 pCi/L (picocuries per liter). Tritium concentrations generally were higher below trench bases (averaging 91,000 pCi/L) than below intertrench sediments (averaging 3,300 pCi/L), and in the subtrench Toulon Member of the Glasford Formation (sand) (averaging 110,000 pCi/L) than in the Hulick Till Member of the Glasford Formation (clayey silt) (averaging 59,000 pCi/L). Average subtrench tritium concentration increased from 28,000 to 100,000 pCi/L during the study period. Within the trench covers, there was a strong seasonal trend in tritium concentrations; the highest concentrations occurred in late summer when soil-moisture contents were at a minimum. Subtrench tritium movement occurred in association with the annual cycle of water movement, as well as independently of the cycle, in apparent response to continuous water movement through the subtrench sand deposits and to the deterioration of trench-waste containers. The increase in concentrations of tritium with increasing distance from the trench bases in the sand unit indicates that water movement through the unit may be more pronounced than indicated by the pressure-head data. Localized, preferential flow paths may have gone undetected by the monitoring instruments used in the study.