http://en.wikipedia.org/wiki/Reentrant_%28computing%29

It means the same section of code can be executing from one user (or process) and a second user can then use that same code again and start executing it at that same time. It promotes some scalability, since you need less resources for that section of code.

If it's not re-entrant, then a second copy of the code must be created and resources assigned for the second user to execute it.