We consider two basic versions of the challenge-response authentication protocol, and exhibit both a method of attack and a simple modification preventing such attacks. We go on to consider three variants of the basic protocols and show that one of them is completely insecure.