Current debate regarding the origin of the Columbia River Basalt Group centers on whether the considerable geochemical diversity in these basalts is the result of derivation from a heterogeneous mantle or of extensive interaction of the magmas with crustal rocks. Compositionally diverse rocks such as the Pomona and Elephant Mountain Members of the Saddle Mountains Formation have identical (87Sr/86Sr ratios (0.7078–0.7080). These radiogenic rocks have nearly identical δ18O values (6.24–6.36) to those of Picture Gorge (6.24), the least radiogenic of the Columbia River Basalts (87Sr/86Sr = 0.7038). This suggests that the mantle beneath the Columbia Plateau is heterogeneous at least with respect to Sr and probably other components. Basalts with low δ18O values (6.19–6.36) can be found in all formations of the Columbia River Basalt, an observation not compatible with models requiring extensive assimilation of upper-crustal rocks to explain isotopic differences between formations. Evidence for crustal contamination for some flow rocks in the Grande Ronde, Wanapum, and Saddle Mountains Basalts does exist. Contamination, however, seems to have affected isolated members of formations rather than formations as a whole. This conclusion is consistent with the results of trace-element modeling, which suggests that the Grande Ronde and Saddle Mountains Basalts were produced as numerous independent magmas from a heterogeneous mantle. The magmas ascended following independent paths, some undergoing crustal contamination, others not.