Grammaticalization of the future tense has been one of the most popular topics in grammaticalization theory. However, little has been said so far about the intermediate stages different constructions (e.g. verbs which formerly meant ‘want’ or ‘go') go through before they finally evolve to mark future temporal reference. I focus on the semantic typology of constructions similar to English about to and going to, which I dub, after Comrie (1976), prospective aspect. I am going to show that (a) there are (at least) three different varieties of prospective aspect, which behave differently with respect to a number of tests; (b) that those three varieties are indeed stages of a grammaticalization into future and (c) this diachronic cline is connected to the general tendency of grammaticalizing items to move up the verbal spine.