The Evje-Iveland pegmatite field in Norway contains pegmatites that are known for their rare scandium mineralization. The petrogenesis of these pegmatites has been debated in the literature for nearly a century. Hypotheses for the origin of the pegmatite-forming melt have included either anatexis of the host amphibolite in vapor-absent conditions, wherein scandium is scavenged from the host amphibolite; or magmatic differentiation, wherein scandium is concentrated through magmatic processes. In order to test the hypothesis that the pegmatite-forming melt was sourced from the host amphibolite, partial melting experiments on the host amphibolite have been performed. These experiments were performed at temperatures ranging from 700 to 1064 °C and pressures between 400 and 550 MPa in a piston-cylinder apparatus. The solidus of the host amphibolite has been determined to be approximately 900 °C at 500 MPa and is significantly higher than the temperature of pegmatite formation. Partial melting of <40% can produce glasses that are broadly granitic in composition and are aluminum- and sodium-rich; however, they are less siliceous than the Evje-Iveland pegmatites. These glasses are also scandium- and REE-poor, and have REE patterns similar to leucosomes in vein-type migmatites, produced at low pressures, but dissimilar to the Evje-Iveland pegmatites. The results of these experiments are thus inconsistent with the hypothesis that the Evje-Iveland pegmatites or, by extension, other rare-element pegmatites, are the result of direct anatexis alone of common metamorphic rock such as amphibolites. It is proposed that the formation of the Evje-Iveland pegmatites is the result of partial melting of a scandium-rich ultramafic or mafic complex, differentiation of that partial melt, and emplacement of that melt into the host amphibolite. Thus, the pegmatite-forming melt may represent the final stages of magmatic differentiation, which is the preferred model for the formation of the Evje-Iveland pegmatites.