The notion of a general block structured programming language and the “most recent” evaluation strategy for such a language are formalized and their relationship investigated. Sufficient conditions for “most recent” correctness are obtained.