Statistical Analysis...

Any statisticians want to tackle this problem?

We have 3 students take an exam in the same room.  When you compare their scantron sheets, they have all the same answers and all the same wrong answers on the 43 multiple choice questions.  There are 1 question deviations between the 3 of them (where they may have gotten one different question right or wrong, but otherwise their exams were identical).

What are the odds of that?  So 2 students scored 23's on their multiple choice section and the 3rd scored a 25.

Here's the issue, how do you pursue this?  Do you nullify all of the scores?  Is this enough evidence to make the claim that cheating took place?  We've decided that if their final scores were not passing, we wouldn't raise the issue, but if they end up passing the exams, what should be done?