We present an implementation of the protocol of Lindell and Pinkas for secure two-party computation which is secure against malicious adversaries [13]. This is the first running system which provides security against malicious adversaries according to rigorous security definition and without using the random oracle model. We ran experiments showing that the protocol is practical. In addition we show that there is little benefit in replacing subcomponents secure in the standard model with those which are only secure in the random oracle model. Throughout we pay particular attention to using the most efficient subcomponents in the protocol, and we select parameters for the encryption schemes, commitments and oblivious transfers which are consistent with a security level equivalent to AES-128.