''Perfect zero-knowledge arguments'' is a cryptographic primitive whic
h allows one polynomial-time player to convince another polynomial-tim
e player of the validity of an NP statement, without revealing any add
itional information (in the information-theoretic sense). Here the sec
urity achieved is on-line: in order to cheat and validate a false theo
rem, the prover must break a cryptographic assumption on-line during t
he conversation, while the verifier cannot find (ever) any information
unconditionally. Despite their practical and theoretical importance,
it was only known how to implement zero-knowledge arguments based on s
pecific algebraic assumptions. In this paper we show a general constru
ction which can be based on any one-way permutation. The result is obt
ained by a construction of an information-theoretic secure bit-commitm
ent protocol. The protocol is efficient (both parties are polynomial t
ime) and can be based on any one-way permutation.