The magnetic exchange interactions in copper(II) mu(2)-azido bridged complexes have been studied using several density functionals and both GTO and STO basis sets. From a methodological point of view, we have taken into proper account nonorthogonality effects in the framework of the broken symmetry approach. A remarkable agreement with experimental data has been obtained using the crystallographic geometry and the new MPW1PW functional. However, modeling of the true ligands by ammonia molecules and complete optimization of the geometry of the isolated complex significantly deteriorate the results. While this can lead to limitations on quantitative studies, general trends and magnetostructural correlations remain very significant. These results are, furthermore, not very sensitive to technical details, like the form of the functional or the type of basis set used.