1 pass PRNG will stop software recovery methods, but will not protect against a detailed analysis in a forensic lab.
7 pass PRNG will give you quite a good protection against hardware recovery, with a probability of 0.99 that every area of the disk will be reoriented once.
33 pass PRNG will very likely stop all hardware recovery attempts, with a probability of 0.99 that every area gets reoriented at least twice.
In all 3 cases, you have to make sure, that your HDD does not contain bad sectors. You can check for bad sectors with specific utilities that you can obtain from the disk manufacturer.
For Samsung this is the hutil.
I'd recommend you download the Ultimate Boot CD, UBCD.
Most drive specific utilities are on that CD, no matter whether you have a Maxtor or a Samsung, you'll likely find the one that applies to your harddrive.
Furthermore, you should only do 1 round PRNG at once. When the first round is finished (you selected 1 round in the menu), then restart your PC and repeat the process x times. Thus you can be sure, that the random noise is not stored in the internal cache and thus only the last pass is written to the drive, although DBAN already checks for this.
Now you might ask, how do I know this 7-pass and 33-pass stuff?
Well, in 2003 a member of one of the federal data protection agencies (yea, we got this cool stuff in Germany
) told a News magazine, that a deletion standard, that is advertised by our Federal Office for IT Security (
www.bsi.de) isn't really secure. The BSI advertises a 3-pass standard without random numbers.
This guy, Roy Pfitzner, wrote an internal paper for his boss at the Agency.
However, this paper was then classified, because in his paper he explained how you could delete data so that even the intelligence services would not get anything, if properly applied.
Nevertheless, it seems that the Data protection agency somehow managed to make this information official.
In October 2004 they published a guide line on secure deletion of data from hard disks, you can find the german version here:
http://www.lfd.m-v.de/informat/magloe/magloe.html
I'm already working on a translation, but I'm still in exams, so you'll have to wait until end of March, unless someone else picks it up.
Maybe I'll ask my local Agency whether they can publish the original paper as well.
greets