A method for detecting malicious code in a stream of data traffic input (400) to a gateway in a data network by monitoring for suspicious data in the stream of data traffic (407). Upon detecting the suspicious data, an attempt is made to disassemble the suspicious data (403) and a treat weight is assigned for each instruction. The attempt to disassemble is initiated at initial instructions each with a different offset within the suspicious portion of data. The threat weights are accumulated respectively for each branch option in the disassembled code (403), producing respectively an accumulated threat weight for each branch option. When the accumulated threat weight exceeds a previously defined threshold level, an alert is generated and/or traffic is blocked from the source of the malicious code.