Using optical rotation to study the triple helix reversion of gelatin in aqueous solutions demonstrates that the reversion is a combination of first-order and second-order kinetic processes. On the basis of this observation, we propose a new two-step mechanism of triple helix formation in polypeptides, that leads to an expression distinct from the one obtained by Flory and Weaver. The rate-limiting step is formation of a two-stranded nucleus, which can be intramolecular (first order) or intermolecular (second order). The triple helix is formed by subsequent wrapping of a third strand onto this nucleus. We estimate the minimum stable helix length and the size of the loop at the end of an intramolecular helix from our kinetics analysis. The new two-step mechanism of triple helix formation is consistent with all existing literature data and allows prediction of concentration and temperature dependencies of helix formation rate.