As NuclearWarhead has pointed out Schwyzer-Debrunner is more modern from a linguist's point of view. Especially in the parts concerned with morphology it has a strongly diachronic approach, which uses also heavily results of Indo-European studies and the like, but also in the syntax parts. The work by Kühner, Blass and Gerth is a few years older and has the more traditional 19th century approach which is more accessible to the non-expert in my opinion. But both grammars are so big and by the mass of material they cover also confusing that you only refer to them if you really are an expert or have a quite exotic question and if you are willing to work through a whole lot of paragraphs, annotations and examples.
(But there isn't anything as comprehensive out there as far as I know.)