Evaluated was an approach to controlling the consolidation of in-place portland cement concrete on the basis of densities obtained with a Troxler 3411 nuclear gage. The gage was used in the backscatter mode on low-slump concrete bridge deck overlays and in both the backscatter and the direct transmission modes on continuously reinforced concrete pavements. The nuclear gage readings were compared to rodded unit weight determinations by ASTM C138. It was found that the nuclear readings compared favorably with the unit weight values. However, a petrographic examination of the cores taken from the deck and pavement indicated the presence of coarse voids in amounts exceeding those found in adequately consolidated concretes. It is concluded that the variability in density resulting from acceptable variations in the grading and amounts of individual ingredients of concrete, including the air content, is of the same order of magnitude or greater than the variability resulting from small but detrimental amounts of large air voids. Consequently, matching nuclear densities with densities obtained by weight per volume measurements does not assure the absence of small but detrimental excesses of large voids. However, the presence of an appreciable amount of honeycombing or excessive air entrainment would be detected by nuclear density measurements.