Windows ME definitely had that reputation. I'll stick up for Windows NT based versions of Windows on stability grounds. But while 95, 98, and 98SE were "mostly

Windows ME definitely had that reputation. I'll stick up for Windows NT based versions of Windows on stability grounds. But while 95, 98, and 98SE were "mostly stable" Windows ME was "mostly unstable" even on good hardware. The 9x based versions of Windows had a lot of problems. And as more LOC were added to Windows 9x it seems like the core design of that range showed that it couldn't scale (since an error in one sub-system "infected" another, due to limited memory protections across boundaries). So often something would crash, but it was actually a sister process which caused it (opps!). Essentially what I am saying is that the 9x range had a "upper complexity limit," so as it became more complex, it inherently became less stable (since more complexity=more bugs=less stability). ME was the upper limit of how complex that model could become. NT based versions of Windows (NT 3.xx, 4.0, 2K, XP, Vista, 7, 8, 8.1, and 10 (and server + embedded editions)) have no such problem since the core design is simply superior (in a large part due to x86 hardware improvements NT takes advantage of). Edit: Just to clarify: "Simply superior" in the final paragraph is contrasting Windows 9x to Windows NT (not OS X). OS X and NT both take advantage of x86's memory protections and both try to restrict kernel hotpatching (which was a massive PITA for 9x/MS Dos). They're both "modern" operating systems relative to what 9x was (and Dos, and other similar systems of that time period).