Petrographic and geochemical studies of basalt glass and related rocks from the Deep-Sea Drilling Project (DSDP) provide a random sampling of magmas which have crystallized to form sea floor basement rocks. The samples range in age fro 155 m.y. to less than 4 m.y. and represent a variety of geologic settings within the three major ocean basins. The majority show the mineralogy, textures, and large-ion lithophile (LIL) element-depleted characteristics of modern basalts from mid-ocean spreading ridges (group I). A smaller but compositionally diverse group show distinctive mineralogy and LIL enrichment (group II). Group II magmas from spreading ridges are very similar to group I magmas in terms of major element variations but are more enriched in LIL elements. Group II magma from seamounts and off-ridge centers tends to be olivine normative and enriched in nomrative plagioclase, while that from aseismic ridges tends to be iron-enriched and quartz normative. The diversity among DSDP rocks is only slightly greater than that shown ata single well-sampled mid-Atlantic ridge (MAR) site (FAMOUS). There are a few unusual Fe- or Ti-rich samples which so far are unique to the Pacific and Indian oceans. Compositional diversity in the Indian Ocean is especially striking. Preliminary petrogenetic modeling suggests that the samples as a whole are not products of a single high-level fractionation sequence. Glasses which project on a pseudobinary cotectic in the Di-An-Fo system may differ significantly in FeO*/MgO, TiO2, and LIL element enrichment at comparable points on the cotectic. These liquids probably represent several compositionally distinct mantle-derived partial melts and their fractionation products. Within the data set, individual fractionation sequences can be defined, in which most major elements are adequately explained by simple crystal-melt equliibria. However, even in these sequences, inferred residual liquids tend to show excess LIL element enrichment which is not yet understood. Most DSDP basement rocks be related back to a spreading ridge environment if due account is taken of apparent 'plume-related' geochemical variations along modern ridges. Unusual LIL element-enriched sea floor basalts may be related to areas such as the one at 45¿N on the MAR where there is little evidence of structural or bathymetric anomalies associated with a 'plume'. We see no evidence for geochemical or mineralogical features diagnostic of variations in spreading rate, nor is there any obvious relation between composition and age.