Cementation in some 400 thin-sections of Pennsylvanian quartzose sandstones of the Eastern Interior, Midcontinent, Michigan, and Appalachian basins was studied petrologically. Of these thin-sections, 120 that showed significant amounts of mineral cement were studied in detail. Parameters measured included (1) amount of secondary silica, (2) the number of enlarged detrital grains that showed detrital outlines, (3) interpenetration of detrital grains and (4) geometry of grain contacts. Amounts of clay matrix and other cements (carbonate, barite) had been determined in an earlier study. The well known inverse relationship between mineral cement and clay matrix holds for these rocks and thus mainly rocks with low clay content were in the group of slides studied in detail. 1 he distribution of silica cement in these rocks is not clearly related to depth of burial, structural position, or stratigraphic horizon. Furthermore, silica cementation is not necessarily related to solution interpenetration of detrital grains. Thus the silica cement, precipitated from solution, may be tied to the chemistry of the original sedimentation environment or to the composition of groundwater solutions passing through the rocks. An evaluation of the chemistry of silica m solution and the solubility of quartz and amorphous silica as related to pH, temperature, and ionic strength of solutions, based on experimental work by Kennedy, Iler, Krauskopf, and others leads to conclusions on the possible chemical environments that might produce silica precipitates. The petrology and knowledge of the general geology of the rocks then enables one to rate the relative probabilities of the several chemical environments that may account for the silica cements actually observed. Although relative time of precipitation can be determined from a petrologic study of the relations between various cements, the placement of the precipitate as “early”, “middle,” or “late” in the diagenetic history of the rock is best done by a combination of petrology and chemistry.