We present high-precision Mg isotope data for most classes of basaltic meteorites including eucrites, mesosiderite silicate clasts, angrites and the ungrouped Northwest Africa (NWA) 2976 measured by pseudo-high-resolution multiple-collector inductively coupled plasma mass spectrometry and utilising improved techniques for chemical purification of Mg. With the exception of the angrites Angra dos Reis, Lewis Cliff (LEW) 86010, NWA 1296 and NWA 2999 and the diogenite Bilanga, which have either been shown to have young ages by other dating techniques or have low Al/Mg ratios, all bulk samples of basaltic meteorites have 26Mg excesses ( δ 26 Mg ∗ = + 0.0135 to +0.0392‰). The 26Mg excesses cannot be explained by analytical artefacts, cosmogenic effects or heterogeneity of initial 26Al/ 27Al, Al/Mg ratios or Mg isotopes in asteroidal parent bodies as compared to Earth or chondrites. The 26Mg excesses record asteroidal melting and formation of basaltic magmas with super-chondritic Al/Mg and confirm that radioactive decay of short-lived 26Al was the primary heat source that melted planetesimals. Model 26Al– 26Mg ages for magmatism on the eucrite/mesosiderite, angrite and NWA 2976 parent bodies are 2.6–3.2, 3.9–4.1 and 3.5 Myr, respectively, after formation of calcium–aluminium-rich inclusions (CAIs). However, the validity of these model ages depends on whether the elevated Al/Mg ratios of basaltic meteorites result from magma ocean evolution on asteroids through fractional crystallisation or directly during partial melting. Mineral isochrons for the angrites Sahara (Sah) 99555 and D’Orbigny, and NWA 2976, yield ages of 5.06 - 0.05 + 0.06 Myr and 4.86 - 0.09 + 0.10 Myr , respectively, after CAI formation. Both isochrons have elevated initial δ 26 Mg ∗ values. Given the brecciated and equilibrated texture of NWA 2976 it is probable that its isochron age and elevated initial δ 26 Mg ∗ ( + 0.0175 ± 0.0034 ‰ ) reflects thermal resetting during an impact event and slow cooling on its parent body. However, in the case of the angrites the marginally elevated initial δ 26 Mg ∗ ( + 0.0068 ± 0.0058 ‰ ) may reflect either δ 26 Mg ∗ ingrowth in a magma ocean prior to eruption and crystallisation or in an older igneous protolith with super-chondritic Al/Mg prior to impact melting and crystallisation of these angrites, or partial internal re-equilibration of Mg isotopes after crystallisation. 26Al– 26Mg model ages and an olivine + pyroxene + whole rock isochron for the angrites Sah 99555 and D’Orbigny are in good agreement with age constraints from 53Mn– 53Cr and 182Hf– 182W short-lived chronometers, suggesting that the 26Al– 26Mg feldspar-controlled isochron ages for these angrites may be compromised by the partial resetting of feldspar Mg isotope systematics. Even when age constraints from the 26Al– 26Mg angrite model ages or the mafic mineral + whole rock isochron are considered, the relative time difference between Sah 99555/D’Orbigny crystallisation and CAI formation cannot be reconciled with Pb–Pb ages for Sah 99555/D’Orbigny and CAIs, which are ca. 1.0 Myr too old (angrites) or too young (CAIs) for reasons that are not clear. This discrepancy might indicate that 26Al was markedly lower (ca. 40%) in the planetesimal- and planet-forming regions of the proto-planetary disc as compared to CAIs, or that CAI Pb–Pb ages may not accurately date CAI formation, which might be better dated by the 182Hf– 182W and 26Al– 26Mg chronometers as 4568.3 ± 0.7 (Burkhardt et al., 2008) and 4568.5 ± 0.3 Ma (herein), respectively, when mapped onto an absolute timescale using Pb–Pb ages for angrites.