Abstract Core-collapse supernovae (CCSNe) are widely accepted to be caused by the explosive death of massive stars with initial masses ≳8 M _⊙. There is, however, a comparatively poor understanding of how properties of the progenitors—mass, metallicity, multiplicity, rotation, etc.—manifest in the resultant CCSN population. Here, we present a minimally biased sample of nearby CCSNe from the All-Sky Automated Survey for Supernovae survey whose host galaxies were observed with integral-field spectroscopy using MUSE at the Very Large Telescope. This data set allows us to analyze the explosion sites of CCSNe within the context of global star formation properties across the host galaxies. We show that the CCSN explosion site oxygen abundance distribution is offset to lower values than the overall H ii region abundance distribution within the host galaxies. We further split the sample at 12 + log 10 ( O / H ) = 8.6 dex and show that within the subsample of low-metallicity host galaxies, the CCSNe unbiasedly trace the star formation with respect to oxygen abundance, while for the subsample of higher-metallicity host galaxies, they preferentially occur in lower-abundance star-forming regions. We estimate the occurrence of CCSNe as a function of oxygen abundance per unit star formation and show that there is a strong decrease as abundance increases. Such a strong and quantified metallicity dependence on CCSN production has not been shown before. Finally, we discuss possible explanations for our result and show that each of these has strong implications not only for our understanding of CCSNe and massive star evolution but also for star formation and galaxy evolution.