Although I guess the quality of the recording of the fingerprint makes a big difference as would the particular method used for determining a match (or decision rule used for declaring a match), is there a rule of thumbprint, so to speak :rolleyes: on how many false matches you might get out of comparing 1,000 prints one to another (I think that makes ~half a million comparisons)?
According to the wikipedia article automated and manual checking have never found two different people with the same prints.
However, computer based methods using hash functions (as used in authentication systems) are less discriminating, with collisions possible within an sample space of several millions.
However, there are those who challenge the validity and objectiveness of fingerprinting - not the theory, but the way it has been applied.
In the last few years fingerprint analysis has undergone a credibility problem. Often an expert will claim positive results on a partial print. The problem is how much of a print is really necessary and does a partial really exclude others. It is not like TV.