Well, you would use 'exec' if you wanted to run an external program as the last step of your Perl script. Control would never return to your Perl script, you'd be finished the program at that point.

You would choose 'system' if you were in the middle of your Perl script, you wanted to run an external program, but you had more Perl code to be run before you were finished. Even if you simply wanted to check to return status of the program you run, you'd need to use system.

In most cases, if you have system vs. exec, system tends to be the one you want to use. If you use exec, you have to trust the external program that it run correctly, there wouldn't be a way from Perl to check that, since the Perl script quits at the exec statement, and gives full control to the external program.

I tried to think of an example of a good use for exec, but every example I could come up with seemed to be better off with system -- largely because I felt it would be best to check that program's return status. However, someone else may be able to come up with an example of where that wouldn't be necessary.
-Eric