Why re-test

Here another question I was asked recently.

Why do we have to re-test some test cases as they worked in the previous cycle?

Imagine you are buying a car. On Top-Gear you heard some very good critics about the car. And you go to the garage. Still you are not that foolish and ask for a test drive. During the ride you step on the accelerator, turn on the radio, open the window to hear the sound of the engine and you are almost convinced.

The sales manager is a very skilled person and tells you that there are more extra's you can choose of. And of course he fooled you and you also want the engine get tuned, a shining outflow and a DVD/navigation system.

After a month you are called that the care is ready and you can pick it up. Don't you use a new test drive to see that the tuning didn't disturb anything else, that everything you liked about the car is still the same?
I would. I would spare me some extra time for the test drive.

In short, you make a test drive when modifications are made to your car, why not perform a regression test on your system to see that everything you liked (no defects) is still working.