Recent reviews and meta-analyses of clinical data have revealed that there is a clear need in objective biomarkers of mental disorders and, in particular, of depression. There is a lot of evidence that EEG coherence can be an important marker of depressive disorders, and can predict response to different antidepressants. The most consistent finding is a decrease of frontal interhemispheric EEG coherence which is observed for most disorders related to depression. Methodological issues are discussed with a particular emphasis on use of statistical classifiers and on coupling EEG coherence with other methods of EEG analysis.