Recently, there have been determined strontium isotopic ratios on granulitic-, eclogitic- and ultra-mafic-nodules originated in the lower crust or in the upper mantle, and on basalts and kimberlites enclosing them. The ratios are largely different between the constituent minerals of nodules and between nodules and their host rocks. This fact suggests that even in the temperature condition of the upper mantle or lower crust isotopic exchange reaction does not take place. Source materials of basaltic magmas in the upper mantle, therefore, are probably composed of minerals with different strontium isotopic ratios. As the constituent minerals are differentially melted in partial melting of rocks to form basaltic magmas, the resulting magmas are probably not in strontium isotopic equilibrium with the bulk of the solid residues, and then the initial strontium isotopic ratios in magma must vary with advancing degree of partial melting. Using the data from nodules as the strontium isotopic ratios of the upper mantle or lower crustal materials, the present authors have attempted to calculate the variations of strontium isotopic compositions of liquids during the partial melting of a mica-bearing garnet-lherzolite (mica+garnet-f-olivine + orthopyroxene + clinopyroxene) and a basic granulite (quartz + K-feldspar + plagioclase + orthopyroxene + clinopyroxene). Liquids produced by 5-20% melting of micabearing garnet-lherzolite have 87Sr/86Sr ratios ranging from about 0.710 to 0.706 and melting more than 20% would yield liquids with lower 87Sr/86Sr ratios ranging from about 0.705 to 0.704. The 87Sr/86Sr ratios of liquids formed by 5-10% melting of basic granulite vary from about 0.712 to 0.708, while melting more than 10% would yield liquids having a relatively limited range of 87Sr/86Sr ratios between 0.709 and 0.707. Based on these results and the results of recent experimental studies on the formation of basaltic and more acid magmas, the present authors have discussed a possibility that the variations of the strontium isotopic composition among the Cenozoic volcanic rocks can be explained by the difference of degree of partial melting of certain source materials similar to the above Iherzolite and basic granulite. Namely, sympathetic variation of 87Sr/86Sr ratios and 87Sr/86Sr ratio in the tholeiitic basalts in the L zone (oceanic ridge and rise, other deep ocean bottom, oceanic island, Type II island arc, large part of Northeast Japan and the Pacific margin of the Cenozoic orogenic belt of American continents) and the alkali basalts in the L zone and H zone (North Island, New Zealand, large part of South-west Japan and Japan Sea side of Northeast Japan and the inner side of the Cenozoic orogenic belt of American continents) may be attributed to the difference of degree of partial melting of certain peridotitic source material. The relatively constant lower 87Sr/86Sr ratios of tholeiitic and calc-alkaline andesites, dacites and rhyolites from the L zone may be explained by the fractional crystallization of tholeiitic basaltic magma. Another possibility is that more acid volcanic rocks such as dacite and rhyolite may be formed by the fractional crystallization of andesitic magma probably derived by direct partial melting of the hydrous upper mantle peridotitic source material. On the other hand, the relatively higher 87Sr/86Sr ratios of tholeiitic and calc-alkaline volcanic rocks in the H zone may possibly be ascribed to the various degree of partial melting of certain basic rocks probably constituting the lower crust or uppermost mantle under the H zone.