Multivariate statistical analysis was applied to examine correlations between reservoir quality and petrology in two data sets from the Middle Jurassic Brent Group. One of the data sets is from relatively shallow depth and has been little affected by chemical diagenesis (Statfjord Nord and Ost Fields; 2.3-2.6 km below the sea floor), while the second data set is from a more deeply buried reservoir having an advanced degree of diagenesis (Huldra Field; 3.6-3.9 km). Much of the total variation in porosity and permeability within each data set (0.02 mD to > 7 D in both sets) can be accounted for by laboratory measurements of parameters mainly related to depositional sand quality, including "shaliness" (represented by bulk-rock alumina/silica ratio), early carbonate cement, feldspar content, and grain size. Despite major differences in the proportions of different sedimentary facies in the two data sets, they have similar ranges of depositional sand quality and therefore probably had similar reservoir quality early in their burial history. Deeper burial diagenesis at Huldra Field has shifted the average of both porosity and permeability to lower values and produced a bimodal permeability distribution, apparently reflecting preferential preservation of permeability in the cleaner sandstones. On the basis of these examples, we outline an approach for unmixing the diagenetic and lithologic components of variation in regional compilations of sandstone porosity-permeability data. The procedure and its consequences are illustrated using a regional compilation of core data from the Brent Group of the northern North Sea.