We analyse two recent computations of type II supernova nucleosynthesis by Woosley & Weaver (1995, hereafter WW95) and Thielemann, Nomoto, & Hashimoto (1996, hereafter TNH96), focusing on the ability to reproduce the observed [Mg/Fe]-ratios in various galaxy types. We show that the yields of oxygen and total metallicity are in good agreement. However, TNH96-models produce more magnesium in the intermediate and less iron in the upper mass range of type II supernovae than WW95-models. To investigate the significance of these discrepancies for chemical evolution, we calculate Simple Stellar Population-yields for both sets of models and different IMF slopes. We conclude that the Mg-yields of WW95 do not suffice to explain the [Mg/Fe] overabundance neither in giant elliptical galaxies and bulges nor in metal-poor stars in the solar neighbourhood and the galactic halo. Calculating the chemical evolution in the solar neighbourhood according to the standard infall-model (e.g. Matteucci & Greggio 1986; Timmes, Woosley, & Weaver 1995; Yoshii, Tsujimoto, & Nomoto 1995), we find that using WW95 and TNH96 nucleosynthesis, the solar magnesium abundance is underestimated by 29 and 7 per cent, respectively.We include the relaxation of the instantaneous mixing approximation in chemical evolution models by splitting the gas component into two different phases. In additional simulations of the chemical evolution in the solar neighbourhood, we discuss various timescales for the mixing of the stellar ejecta with the interstellar medium. We find that a delay of the order of 10 8 years leads to a better fit of the observational data in the [Mg/Fe]-[Fe/H] diagram without destroying the agreement with solar element abundances and the age-metallicity relation.