An article on MacFixit this morning tells of a test carried out using the OS X 10.2.6 incremental (6MB) and combo (86MB) updaters from Apple.

The author claims that there 2800 files that are different after installing each on two identical copies of 10.2.5. From this information, he theorizes that the combo updater should always be used.

I'm not sure.

I haven't run the test myself (which I will do), but my guess is that they aren't using the right test. I have noticed that running the updaters will commonly change file dates even if their contents do not change. Because of this, many differencing programs determine that the files have "changed", however, the information in them is identical, and therefore it is a waste of time to use the combo updater. I will MD5 checksum the files, which should give a better assessment of this.

However, if you believe your system is unstable after an OS update, I do agree that re-running the update using the combo updater would be a good idea.