Abstract: Fault induction attack assumes that the attacker is able to induct faults into the cryptographic device (smartcard) and make it output incorrect encryption results. The attacker exploits the correct and incorrect encryption results to disclose the secret information (key) hidden in the cryptographic device. A method of the fault induction attack on RC4 algorithm is brought forward. The simulations show that one attack may find the values of 3 positions in the initial state of RC4 and continuous attacks can recover the whole initial state of RC4 with a considerable probability (more than 1/2). About O(2 16) key stream bytes at most are needed to recover the whole initial state of RC4 after about O(2 16) fault inductions at most.