Abstract Non-malleable code is an encoding scheme that is useful in situations where traditional error correction or detection is impossible to achieve. It ensures with high probability that decoded message is either completely unrelated or the original one, when tampering has no effect. Usually, standard version of non-malleable codes provide security against one time tampering attack. Block ciphers are successfully employed in the construction of non-malleable codes. Such construction fails to provide security when an adversary tampers the codeword more than once. Continuously non-malleable codes further allow an attacker to tamper the message for polynomial number of times. In this work, we propose continuous version of non-malleable codes from block ciphers in split-state model. Our construction provides security against polynomial number of tampering attacks and it preserves non-malleability. When the tampering experiment triggers self-destruct, the security of continuously non-malleable code reduces to security of the underlying leakage resilient storage.