I havent done it for car audio yet, but with my HT setup, what i do is pick a decent volume level, play a fequency depending on what you're testing, like maybe 25-30hz to start for the sub, and adjust the vol somehow while just watching the excusion, to make sure you're not going to blow it.... and keep the vol at a safe level, whatever you think might not blow the sub... now it really depends why you're playing test tones... if you're just testing the subs power, then just watch it, thats all, but if you're doing some measuring, the vol dosent have to be cranked loud as the sub/amps will go, since you're just measuring the frequency responses loudness... so you basically find a safe volume level and leave it there, dont touch it, and test the tones from 20hz and up to whatever, and meter each one with whatever you got, eg, radio shack SPL meter... (wich isnt 100% accurate, but there's correction values floating around) and write these numbers down... get some test tones that test every frequency in you're range and not skip any, like 20hz, 21,22,23,24,25, etc... this can take a while though sometimes...hehe.
Once you write em down, make a graph with em and you can see the response curve for the sub/car acoustics/Box....
If you're sub in its best box, and the car is dampend its best, and the response has lots of dips or spikes at certian freqs, you could always get a Parametric EQ to flatten out the response so all the frequencies play at the same level for great SQ....