There are various solutions around the net for this problem, assuming one does not want to replace the mouse:
1) Of course, the definitive one is to substitute the misbehaving microswitch. Since the Omron D2FC-F-7N microswitches are available on e-Bay for under US$5 a pair, that's the solution I favor, and have used to keep my three Intellimouse Optical v 1.1 mice working (I have already substituted the left upper button microswitch twice in each of them, and still have not seen the problem happen with any of the other buttons).
2) Then, there is the leaf-spring remove-retension-reinstall solution, which is pretty well described here (how to fix the 'double-click' problem on your G7 mouse), but which I was never able to do (all I accomplished was to make the already bad microswitch completely unresponsive). If one manages to perform it correctly, it should prolong the life of the microswitch somewhat.
3) Finally, there's an interim software solution, which really helps overcome the double-clicking on single-clicking issue, in its initial stages, although it gets innefective as the spring gets worse. This is the freeware program MouseFix (direct download), by Daniel Jackson. His MouseFix page (please scroll way down) is worthy of reading, and also offers the program's source code.
The usual way of running MouseFix is simply to add a link it to the Startup folder in the Start Menu... however, since MouseFix hooks one of the mouse events, this leads to the cursor becoming unresponsive at times, due to it being run at Normal priority. My solution is to set it to High or, even better, RealTime priority. To do this automatically on startup, edit the Target field in the link to MouseFix, so that it becomes:
C:\WINDOWS\system32\cmd.exe /c start "" /B /REALTIME "C:\Program Files\MouseFix\MouseFix.exe"if your copy of MouseFix is at C:\Program Files\MouseFix\, or adjust the drive and path as needed, and set the Run field to Minimized. This'll start MouseFix with RealTime priority, eliminating all unresponsiveness.
BTW, there's KB266738 (and its 2003 version), but IMO that MS document induces one to check many things pointlessly, inspiring false hopes, before coming round to realize it, in fact, is a hardware problem. There's also a still open thead, from 2005, at neowin about it, and I cite it here as one of sources of part of the info above, and for its historical value.