Several Early Proterozoic Hearst–Matachewan (2.454 Ga), Kapuskasing (2.14 Ga), and Preissac (2.04 Ga) dykes were emplaced within the Archean crust surrounding the Kapuskasing structural zone (KSZ). The dykes are composed of moderately to highly fractionated tholeiitic basalts (Mg number = 24–55) that exhibit trace-element characteristics similar to those of intraplate basaltic magmas or ocean–island basalts (e.g., Zr/Nb = 6–21, Zr/Y = 2–5, high TiO2 = 0.9–3.2 wt.%, and (Fe2O3)t = 12.4–18.7 wt.%). Their initial Nd isotopic compositions display a range of depleted to enriched values that are negatively correlated with the degree of light rare-earth element enrichment. We evaluate two models for the origin of these dykes: (i) The basaltic parental magmas were derived from two distinct sources, an isotopically depleted asthenospheric mantle (εNd = +4 and La/Sm = 2.7) and an isotopically enriched lithospheric(?) mantle (εNd = −4 to−8 and La/Sm = 5.1). The magmas subsequently underwent mixing and fractionation during ascent in the mantle or the lower crust. (ii) The parental magmas originated from a homogeneous Nd isotopically depleted asthenospheric mantle but later assimilated a substantial amount of Archean crustal material upon fractionation and ascent in the lower crust. Results derived for the latter model preclude any participation of the exposed crustal rocks in the KSZ, and the assimilation and fractional crystallization (AFC) model remains a viable hypothesis only if the parental magmas assimilated an older and perhaps more isotopically enriched crust than that represented in the KSZ.