In nuclear astrophysics, optical model potentials play an important role, both in the nucleosynthesis models, and in the interpretation of astrophysics related nuclear physics measurements. The challenge of nuclear astrophysics resides in the fact that it involves many nuclei far from the stability line, implying than very few (if any) experimental results are available for these nuclei. The answer to this challenge is a heavy reliance on microscopic optical models with solid microscopic physics foundations that can predict the relevant physical quantities with good accuracy. This use of microscopic information limits the likelihood of the model failing spectacularly (except if some essential physics was omitted in the modeling) when extrapolating away from the stability line, in opposition to phenomenological models which are only suited for interpolation between measured data points and not for extrapolating towards unexplored areas of the chart of the nuclides.We will show how these microscopic optical models are built, how they link to our present knowledge of nuclear structure, and how they affect predictions of nuclear astrophysics models and the interpretation of some key nuclear physics measurements for astrophysics. © 2004 American Institute of Physics