Semiconductivity (SC) and ferromagnetism (FM) are an unlikely couple, each having quite different desires in regard of the electronic band structure (High density of states at the Fermi level for FM, but low or moderate for SC). In the search for materials that make this difficult relationship work, electronic structure theory has played a prominent role, predicting new diluted magnetic semiconductor (DMS) materials, offering explanations and suggesting models for the underlying physical mechanisms. However, treacherous pitfalls need to be avoided along the path leading from a density functional calculation to a realistic model for DMS. In this talk, I will highlight examples how such calculations can produce “false positives”, but also offer solutions that allows to use density functional methods as powerful tools to describe magnetism in semiconductors by an atomistic electronic structure theory rather than a model Hamiltonian.