ABSTRACT The evolutionary paths taken by massive stars with M ≳ 60 M⊙ remain substantially uncertain. They begin their lives as main-sequence (MS) O stars. Depending on their masses, rotation rates, and metallicities, they can then encounter a wide range of evolutionary states with an equally broad set of possible surface compositions and spectral classifications. We present a new grid of calculations for the evolution of such stars that covers a broad range in mass, M/M⊙ = 60–150, rotation rate, $v/v_{\rm crit} = 0$–0.6, metallicity, [Fe/H] = −4 to 0, and α-element enhancement, [α/Fe] = 0–0.4. We show that rotating stars undergo rotationally induced dredge-up of nucleosynthetic products, mostly He and N, to their surfaces while still on the MS. Non-rotating metal-rich stars also reveal the products of nucleosynthesis on their surfaces because even modest amounts of mass-loss expose their ‘fossil’ convective cores: regions that are no longer convective, but were part of the convective core at an early stage in the star’s evolution. Thus, surface enhancement of He and N is expected for rotating stars at all metallicities, and for non-rotating stars if they are relatively metal-rich. We calculate a stellar atmosphere for a representative model from our grid, properly accounting for He and N enhancement, and show that the resulting spectrum provides a good match to observed WNL stars, strongly suggesting that the physical mechanisms we have identified are the ultimate cause of the WNL phase.