Measurement of the fluid-content distribution at sites contaminated by immiscible fluids, including crude oil, is needed to better understand the movement of these fluids in the subsurface and to provide data to calibrate and verify numerical models and geophysical methods. A laboratory method was used to quantify the fluid contents of 146 core sections retrieved from boreholes aligned along a 120-m longitudinal transect at a crude-oil spill site near Bemidji, Minnesota, U.S.A. The 47-mm-diameter, minimally disturbed cores spanned a 4-m vertical interval contaminated by oil. Cores were frozen on site in a dry ice-alcohol bath to prevent redistribution and loss of fluids while sectioning the cores. We gravimetrically determined oil and water contents using a two-step method: (1) samples were slurried and the oil was removed by absorption onto strips of hydrophobic porous polyethylene (PPE); and (2) the samples were oven-dried to remove the water. The resulting data show sharp vertical gradients in the water and oil contents and a clearly defined oil body. The subsurface distribution is complex and appears to be influenced by sediment heterogeneities and water-table fluctuations. The center of the oil body has depressed the water-saturated zone boundary, and the oil is migrating laterally within the capillary fringe. The oil contents are as high as 0.3cm³cm⁻³, which indicates that oil is probably still mobile 10 years after the spill occurred. The thickness of oil measured in wells suggests that accumulated thickness in wells is a poor indicator of the actual distribution of oil in the subsurface. Several possible sources of error are identified with the field and laboratory methods. An error analysis indicates that adsorption of water and sediment into the PPE adds as much as 4% to the measured oil masses and that uncertainties in the calculated sample volume and the assumed oil density introduce an additional ±3% error when the masses are converted to fluid contents.