In my tests I would say creating a 'solid archive' in WinRAR gives you the best results (this counts for both, version 4 and 5) but the downside of it is extracting a single file (like an *.nfo) from the archive takes way longer. But this does not guarantee that the archive size is smaller in every scenario.
If the developers of WinRAR say that version 5 creates smaller archives than version 4 then this could be the case for most of the scenarios like compressing text files, compressing word documents and so on. You know this only after the compression has finished. The problem is the data itself so you cannot foresee which algorithm or what combination you should use to achieve the smallest archive. Lempel–Ziv–Welch algorithm, Run-length encoding etc that are maybe used by WinRAR and WinZip). I went with WinRar as follows, this does not require you to create an sfx file, everything can be created via the gui: Select files, right click and select Add to Archive Use Browse. You can also combine some of them in sequence to compress the data even more (I do not mean WinRAR and WinZip but i.e. Each of them is more suitable for one or for another case. There are many compression algorithms out there.
