Recognition of the magnitude and tectonic significance of oceanic volcanism has recently stimulated intensive investigation of the volcanic rocks and structures of the sea floor. An adequate interpretation of these features depends on a thorough understanding of submarine volcanic mechanisms. The pressure of deep water restricts the amount of vesiculation of eruptive magmas so that explosive activity is suppressed. Pillow lavas are probably shorter but thicker than subaerial flows of the same composition. Fragmental hyaloclastites are also important, but seem to be most common at relatively shallow depths. Most lavas and hyaloclastites are basaltic, but more siliceous pyroclastic deposits may be common in shallow seas of the continental shelves.Circumstantial evidence indicates that most abyssal basalts of the ocean floor were erupted at spreading centers and have been transported laterally. The rate of eruption of lavas is probably related to the mechanism of dike intrusion into a spreading center. Slow spreading rates result in less vertical transfer of heat, so that only relatively thick dikes reach the surface. Thinner dikes are chilled before erupting and cause a differential dilation of the crust that is reflected in shallow normal faulting. With increasing spreading rates, wall rocks are maintained at higher temperatures, and almost all dikes reach the surface to produce lava flows and smoother topography.Seamounts may begin to grow at ridge crests and continue as the volcano moves outward, but many must begin their activity far from the spreading center. Most of the submarine structure of seamounts must consist of pillow lavas with increasing amounts of hyaloclastites in the upper portion. Injection of dikes and sills adds substantially to the volume of the structure.Most abyssal basalts erupted at ridge crests are tholeiites, commonly with relatively high alumina contents. Silica‐deficient alkaline basalts become more common outward, especially at the crests of seamounts. The ultramafic rocks dredged from the mid‐Atlantic ridge fall into two classes. Some are clearly cumulate rocks related to associated gabbro and differentiated basalt. Others are strongly depleted in low‐melting components and appear to have been emplaced along fault zones; they may be derived from a residual layer from which basalt was extracted during an earlier episode of partial melting. It is difficult to find conclusive evidence that oceanic volcanism was as intense in pre‐Mesozoic time as it is postulated to b