Previous Sr-Nd isotopic models of two-component crust-mantle mixing (based on an apparently simple hyperbolic ϵNd-Sri array) have failed to account for all the geochemical and isotopic features of granitoids from the Lachlan fold belt, eastern Australia. Here we show that a three-component mixture of mantle-derived magma and two contrasting crustal components successfully explains the isotopic compositions of the granitoids. New isotopic data from the Moruya granitoids of the fold belt reveal that a separate isotopic trend exists at the most primitive end of the ϵNd-Sri array, defined by a depleted mantle component (ϵNd = +8.0; Sri = 0.7032) and a mafic crustal component (ϵNd = +2.7; Sri = 0.7049). This crustal component has isotopic characteristics similar to Cambrian greenstones within the fold belt, and appears to be widespread in the lower crust. After also considering data from the ubiquitous Ordovician turbidites which host the granitoids (ϵNd = −9.61 to −10.64; Sri = 0.7204 to 0.7273), three-component isotopic mixing curves are constructed that enclose all the granitoids of the fold belt, indicating that contributions from the Ordovician turbidites, Cambrian greenstones, and a depleted mantle component could produce the isotopic characteristics of the granitoids. Important implications of the model are that (1) S- and I-type granites appear to be mixtures of, rather than unique products from, contrasting sources and so Lachlan granitoids do not directly “image” their source-rock compositions; (2) the chemical variation lines defining granitoid suites are neither mixing lines nor lines reflecting restite separation; (3) previously inferred Precambrian continental basement beneath the fold belt is unnecessary; and (4) Lachlan granitoids may have formed in an environment related to subduction, where arc magmatism promoted crustal fusion and crust-mantle mixing.