Temporal information in natural language texts is an increasingly important component to the understanding of those texts. This part of ISO 24617, SemAF-Time, specifies a formalized XML-based markup language called ISO-TimeML, with a systematic way to extract and represent temporal information, as well as to facilitate the exchange of temporal information, both between operational language processing systems and between different temporal representation schemes. The use of guidelines for temporal annotation has been fully attested with examples from the TimeBank corpus, a collection of 183 documents that have been annotated by TimeML before the current version of ISO-TimeML was formulated.