The Hyde Park landfill is a 15-acre (6.1 ha) chemical waste disposal site located north of Niagara Falls, New York. Underlying the site in descending order are: (1) low-permeability glacial till and lacustrine deposits; (2) a moderately permeable fractured rock aquifer - the Lockport Dolomite; and (3) a low-permeability unit - the Rochester Shale. The site is bounded on three sides by groundwater drains; the Niagara River gorge, the Niagara Power Project canal, and the Niagara Power Project buried conduits. The mechanism by which groundwater moves through fractured rocks underlying a hazardous waste site was investigated using a digital simulation approach. Three hypotheses were tested related to flow in the fractured rocks underlying Hyde Park landfill. For this purpose we used a Galerkin finite-element approximation to solve a saturated-unsaturated flow equation. A primary focus was to investigate anisotropy in the Lockport Dolomite, that is the effectiveness of horizontal (bedding) joints vs. vertical joints as water-transmitting openings. Three hydrogeologic scenarios were set up - each with prescribed limits on the hydrologic parameters. Scenario 1 specified strongly anisotropic conditions in the Lockport Dolomite (horizontal hydraulic conductivity along bedding joints exceeds vertical conductivity by 2-3 orders of magnitude), uniform areal recharge (5 in. yr.-1 or 12.7 cm yr.-1) except at the landfill where there is no recharge, and no flow through the base of the Rochester Shale. Scenario 2 also specified strongly anisotropic conditions in the Lockport; however, areal recharge was 6 in. yr.-1 (15.2 cm yr.-1) except at the landfill where the recharge was 2 in. yr.-1 (5.1 cm yr.-1), and outflow from the Rochester occurred. Scenario 3 specified isotropic conditions (that is, permeability along horizontal and vertical joints is the same in the Lockport Dolomite), recharge rates were the same as in scenario 2 and outflow through Rochester occurred. Scenario 2 provided the closest agreement between the simulated and measured heads while scenario 3 provided the poorest agreement. Among the three scenarios tested, scenario 2 (with strongly anisotropic conditions in the Lockport Dolomite with added recharge through the landfill cap and limited flow through the Rocherster Shale) is considered the most realistic hydrogeologic model. Based on simulation with the hydrogeologic parameters of scenario 2, groundwater flow near the Hyde Park site can be summarized as follows: 1. (1) Specific discharge (Darcy velocity) ranges from ???0.01 to 0.1 ft. day-1 (0.003 to 0.03 m day-1) in the upper unit of the Lockport Dolomite to slightly more than 0.0001 ft. day-1 (0.00003 m day-1) in the Rochester Shale. Real velocities are highest in the upper unit of the Lockport, ranging from ???1 to 5 ft. day-1 (0.3 to 1.5 m day-1) if the average effective porosity is assumed to be 0.02. 2. (2) A groundwater divide exists east of the landfill, indicating that all groundwater originating near or flowing beneath the landfill will flow toward and discharge in the gorge. 3. (3) Highest flow velocities (and presumably greatest potential for transporting chemical contaminants) occur in the upper unit of the Lockport and part of the lower unit of the Lockport Dolomite between the landfill and the gorge. The average time required for groundwater to move from the landfill to the discharge points at the gorge along selected flow paths in the Lockport Dolomite is estimated to be 5-6 yr. ?? 1984.