Definitions of "Trojan horse" and related terms have varied widely over the years, but common usage in 2008 generally refers to software that performs a legitimate function, but also contains malicious code.

Almost any malicious code can be called a Trojan horse, since the author of malicious code needs to disguise it somehow so that it will be invoked by a nonmalicious user (unless the author means also to invoke the code, in which case he or she presumably already possesses the authorization to perform the intended sabotage). A Trojan horse that replicates itself by copying its code into other program files (see case MA1) is commonly referred to as a virus. One that replicates itself by creating new processes or files to contain its code, instead of modifying existing storage entities, is often called a worm. Denning provides a general discussion of these terms; differences of opinion about the term applicable to a particular flaw or its exploitations sometimes occur.

Potentially malicious dynamic code compiled at runtime can conceal any number of attacks that will not appear in the baseline. The use of dynamically compiled code could also allow the injection of attacks on post-deployed applications.