Fluvial, floodplain and lake strata of the Green River Formation (Eocene) occur within an exhumed oil reservoir exposed in a quarry near Sunnyside, Utah. Strata in the quarry highwalls define three, asymmetrical, 15- to 20-m thick, base-level-rise genetic sequences arranged in a long-term base-level-rise (landward stepping) stacking pattern. Variable intensity of oil stain on rock surfaces is a qualitative measure of pore volumes, as all permeable facies are fully saturated with oil. Visual estimates of oil-stain intensity, combined with petrophysical measurements and petrographic analysis of the different facies, were used to define fluid-flow compartments and their boundaries. Strata and facies that functioned as fluid-flow conduits, retardants and barriers were mapped on photomosaics of the quarry highwall. Three separate fluid-flow compartments coincide with the three genetic sequences. Amalgamated fluvial sandstones at the base of each genetic sequence functioned as flow units of varying permeability and degree of interconnectedness. Laterally continuous floodplain and/or lacustrine mudstones, which cap each genetic sequence, entirely lack oil in matrix porosity and functioned as fluid-flow barriers and compartment boundaries. Petrophysical properties of specific sedimentary facies are sensitive to stratigraphic position at three spatial scales, even though the sedimentary facies appear nearly identical. At the long-term scale, porosity and permeability of the same facies (trough cross-stratified sandstone is the most common) in channel sandstones of the three genetic sequences decrease in stratigraphic succession. Within each genetic sequence, porosity and permeability are highest at the base and decrease quasilinearly to the top. Using oil-stain intensity as a proxy, porosity and permeability generally decrease from base to top of each scour-based channel macroform. Petrophysical variations coincide with subtle variations in grain size and trough cross-stratification set thickness within otherwise indistinguishable sedimentary facies. These results demonstrate that conventional crossplots of porosity/permeability versus sedimentary facies are unnecessarily broad and imprecise. When such petrophysical data are plotted in a stratigraphic context, porosity and permeability values have significantly reduced scatter. Porosity and permeability measurements and predictions of each sedimentary facies should be made from a stratigraphic perspective. From our observations of variations in intensity of oil stain, homogeneity of fluid flow may not be equated directly with facies homogeneity. At one extreme of an apparent continuum, fluid-flow pathways are tortuous and extremely variable within homogeneous, high permeability, amalgamated channel sandstones. Sweep efficiencies may be low in these cases. At an intermediate position in the continuum, increased diversity of sedimentary facies and stratigraphic variability usually cause sufficient stratigraphic separation of permeable and impermeable strata such that fluid-flow pathways are more confined and have a reduced tortuosity. Sweep efficiencies may be high in these cases. At the other extreme of the continuum, where diversity of sedimentary facies and stratigraphic variability is very high, stratigraphic units are discontinuous and restricted in area. In such cases, fluid-flow pathways are not laterally connected, and sweep efficiencies would be low.