Hypotheses to explain the formation of native-collagen fibrils from tropocollagen macromolecules have been considered and their inconsistencies examined. A new model for the tropocollagen macromolecule is suggested and a new mechanism for its assembly into native-collagen fibrils is offered. It is postulated that the tropocollagen macromolecule is divided into 5 main bonding zones separated by 4 main non-bonding zones. The macromolecules are assembled into native fibrils by allowing a main bonding zone on a particular macromolecule an initial random choice in combining, in a structurally complementry manner, with a main bonding zone on another macromolecule. The evidence for the proposals arises from the use of negative-staining techniques and high-resolution electron microscopy. The work described includes a more detailed discussion of some preliminary findings reported earlier (Grant, Horne, & Cox, 1965b).