The Variscan high‐grade metamorphic basement of northern Sardinia and southern Corsica record lower Carboniferous anatexis related to post‐collisional decompression of the orogen. Migmatites exposed in the Punta Bianca locality (Italy) consist of quartz + biotite + plagioclase + K‐feldspar orthogneisses, garnet and cordierite‐bearing diatexite and metatexites, derived from metasediments. Field evidence, petrographic observations, ELA‐ICP‐MS zircon and monazite dating and pseudosection modelling suggest that anatexis was apparently episodic involving two main stages of partial melting. Using pseudosection modelling, we infer that the first stage of partial melting is in the upper amphibolite facies (~0.45 GPa at ~740°C). Cordierite overgrowths replacing sillimanite, combined with the composition of plagioclase and K‐feldspar, suggest decompression followed cooling below the solidus at low pressures of ~0.3 GPa. The age of the first anatectic event is not precisely constrained because of extensive resetting of the isotopic systems during the second melting stage, yet few zircons preserve a lower Carboniferous age which is consistent with the regional dataset. This lower Carboniferous migmatitic fabric is offset by a network of pseudotachylyte‐bearing faults suggestive of cooling to greenschist facies conditions. Garnet/cordierite‐bearing diatexites incorporate fragments of pseudotachylite‐bearing orthogneiss and metatexites. Pseudosection modelling indicates nearly isobaric re‐heating up to ~750°C, followed by further cooling below the solidus. The inferred P–T path is consistent with decompression and cooling of the Variscan crust through post‐collisional extension and collapse of the thickened orogenic crust, followed by nearly isobaric re‐heating at low pressures (~0.3 GPa) yielding to a second melting stage under LP‐HT conditions. U/Th‐Pb monazite ages for diatexite migmatites indicate an upper bound of 310–316 Ma for the second melting stage, suggesting that the second melting stage is coincident with the regional phase of crustal shearing. The cause of the high geothermal gradient required for re‐heating during the second melting stage is unknown but likely requires some heat source that was probably related to dissipation of mechanical work within crustal‐scale shear zones. According to this interpretation, some upper Carboniferous peraluminous granite precursors of the Corsica–Sardinia Batholith could be the outcome rather than the cause of the late‐Variscan high‐T metamorphism.